bachelor thesis topics machine learning

The Future of AI Research: 20 Thesis Ideas for Undergraduate Students in Machine Learning and Deep Learning for 2023!

A comprehensive guide for crafting an original and innovative thesis in the field of ai..

By Aarafat Islam on 2023-01-11

“The beauty of machine learning is that it can be applied to any problem you want to solve, as long as you can provide the computer with enough examples.” — Andrew Ng

This article provides a list of 20 potential thesis ideas for an undergraduate program in machine learning and deep learning in 2023. Each thesis idea includes an  introduction , which presents a brief overview of the topic and the  research objectives . The ideas provided are related to different areas of machine learning and deep learning, such as computer vision, natural language processing, robotics, finance, drug discovery, and more. The article also includes explanations, examples, and conclusions for each thesis idea, which can help guide the research and provide a clear understanding of the potential contributions and outcomes of the proposed research. The article also emphasized the importance of originality and the need for proper citation in order to avoid plagiarism.

1. Investigating the use of Generative Adversarial Networks (GANs) in medical imaging:  A deep learning approach to improve the accuracy of medical diagnoses.

Introduction:  Medical imaging is an important tool in the diagnosis and treatment of various medical conditions. However, accurately interpreting medical images can be challenging, especially for less experienced doctors. This thesis aims to explore the use of GANs in medical imaging, in order to improve the accuracy of medical diagnoses.

2. Exploring the use of deep learning in natural language generation (NLG): An analysis of the current state-of-the-art and future potential.

Introduction:  Natural language generation is an important field in natural language processing (NLP) that deals with creating human-like text automatically. Deep learning has shown promising results in NLP tasks such as machine translation, sentiment analysis, and question-answering. This thesis aims to explore the use of deep learning in NLG and analyze the current state-of-the-art models, as well as potential future developments.

3. Development and evaluation of deep reinforcement learning (RL) for robotic navigation and control.

Introduction:  Robotic navigation and control are challenging tasks, which require a high degree of intelligence and adaptability. Deep RL has shown promising results in various robotics tasks, such as robotic arm control, autonomous navigation, and manipulation. This thesis aims to develop and evaluate a deep RL-based approach for robotic navigation and control and evaluate its performance in various environments and tasks.

4. Investigating the use of deep learning for drug discovery and development.

Introduction:  Drug discovery and development is a time-consuming and expensive process, which often involves high failure rates. Deep learning has been used to improve various tasks in bioinformatics and biotechnology, such as protein structure prediction and gene expression analysis. This thesis aims to investigate the use of deep learning for drug discovery and development and examine its potential to improve the efficiency and accuracy of the drug development process.

5. Comparison of deep learning and traditional machine learning methods for anomaly detection in time series data.

Introduction:  Anomaly detection in time series data is a challenging task, which is important in various fields such as finance, healthcare, and manufacturing. Deep learning methods have been used to improve anomaly detection in time series data, while traditional machine learning methods have been widely used as well. This thesis aims to compare deep learning and traditional machine learning methods for anomaly detection in time series data and examine their respective strengths and weaknesses.

Photo by  Joanna Kosinska  on  Unsplash

6. Use of deep transfer learning in speech recognition and synthesis.

Introduction:  Speech recognition and synthesis are areas of natural language processing that focus on converting spoken language to text and vice versa. Transfer learning has been widely used in deep learning-based speech recognition and synthesis systems to improve their performance by reusing the features learned from other tasks. This thesis aims to investigate the use of transfer learning in speech recognition and synthesis and how it improves the performance of the system in comparison to traditional methods.

7. The use of deep learning for financial prediction.

Introduction:  Financial prediction is a challenging task that requires a high degree of intelligence and adaptability, especially in the field of stock market prediction. Deep learning has shown promising results in various financial prediction tasks, such as stock price prediction and credit risk analysis. This thesis aims to investigate the use of deep learning for financial prediction and examine its potential to improve the accuracy of financial forecasting.

8. Investigating the use of deep learning for computer vision in agriculture.

Introduction:  Computer vision has the potential to revolutionize the field of agriculture by improving crop monitoring, precision farming, and yield prediction. Deep learning has been used to improve various computer vision tasks, such as object detection, semantic segmentation, and image classification. This thesis aims to investigate the use of deep learning for computer vision in agriculture and examine its potential to improve the efficiency and accuracy of crop monitoring and precision farming.

9. Development and evaluation of deep learning models for generative design in engineering and architecture.

Introduction:  Generative design is a powerful tool in engineering and architecture that can help optimize designs and reduce human error. Deep learning has been used to improve various generative design tasks, such as design optimization and form generation. This thesis aims to develop and evaluate deep learning models for generative design in engineering and architecture and examine their potential to improve the efficiency and accuracy of the design process.

10. Investigating the use of deep learning for natural language understanding.

Introduction:  Natural language understanding is a complex task of natural language processing that involves extracting meaning from text. Deep learning has been used to improve various NLP tasks, such as machine translation, sentiment analysis, and question-answering. This thesis aims to investigate the use of deep learning for natural language understanding and examine its potential to improve the efficiency and accuracy of natural language understanding systems.

Photo by  UX Indonesia  on  Unsplash

11. Comparing deep learning and traditional machine learning methods for image compression.

Introduction:  Image compression is an important task in image processing and computer vision. It enables faster data transmission and storage of image files. Deep learning methods have been used to improve image compression, while traditional machine learning methods have been widely used as well. This thesis aims to compare deep learning and traditional machine learning methods for image compression and examine their respective strengths and weaknesses.

12. Using deep learning for sentiment analysis in social media.

Introduction:  Sentiment analysis in social media is an important task that can help businesses and organizations understand their customers’ opinions and feedback. Deep learning has been used to improve sentiment analysis in social media, by training models on large datasets of social media text. This thesis aims to use deep learning for sentiment analysis in social media, and evaluate its performance against traditional machine learning methods.

13. Investigating the use of deep learning for image generation.

Introduction:  Image generation is a task in computer vision that involves creating new images from scratch or modifying existing images. Deep learning has been used to improve various image generation tasks, such as super-resolution, style transfer, and face generation. This thesis aims to investigate the use of deep learning for image generation and examine its potential to improve the quality and diversity of generated images.

14. Development and evaluation of deep learning models for anomaly detection in cybersecurity.

Introduction:  Anomaly detection in cybersecurity is an important task that can help detect and prevent cyber-attacks. Deep learning has been used to improve various anomaly detection tasks, such as intrusion detection and malware detection. This thesis aims to develop and evaluate deep learning models for anomaly detection in cybersecurity and examine their potential to improve the efficiency and accuracy of cybersecurity systems.

15. Investigating the use of deep learning for natural language summarization.

Introduction:  Natural language summarization is an important task in natural language processing that involves creating a condensed version of a text that preserves its main meaning. Deep learning has been used to improve various natural language summarization tasks, such as document summarization and headline generation. This thesis aims to investigate the use of deep learning for natural language summarization and examine its potential to improve the efficiency and accuracy of natural language summarization systems.

Photo by  Windows  on  Unsplash

16. Development and evaluation of deep learning models for facial expression recognition.

Introduction:  Facial expression recognition is an important task in computer vision and has many practical applications, such as human-computer interaction, emotion recognition, and psychological studies. Deep learning has been used to improve facial expression recognition, by training models on large datasets of images. This thesis aims to develop and evaluate deep learning models for facial expression recognition and examine their performance against traditional machine learning methods.

17. Investigating the use of deep learning for generative models in music and audio.

Introduction:  Music and audio synthesis is an important task in audio processing, which has many practical applications, such as music generation and speech synthesis. Deep learning has been used to improve generative models for music and audio, by training models on large datasets of audio data. This thesis aims to investigate the use of deep learning for generative models in music and audio and examine its potential to improve the quality and diversity of generated audio.

18. Study the comparison of deep learning models with traditional algorithms for anomaly detection in network traffic.

Introduction:  Anomaly detection in network traffic is an important task that can help detect and prevent cyber-attacks. Deep learning models have been used for this task, and traditional methods such as clustering and rule-based systems are widely used as well. This thesis aims to compare deep learning models with traditional algorithms for anomaly detection in network traffic and analyze the trade-offs between the models in terms of accuracy and scalability.

19. Investigating the use of deep learning for improving recommender systems.

Introduction:  Recommender systems are widely used in many applications such as online shopping, music streaming, and movie streaming. Deep learning has been used to improve the performance of recommender systems, by training models on large datasets of user-item interactions. This thesis aims to investigate the use of deep learning for improving recommender systems and compare its performance with traditional content-based and collaborative filtering approaches.

20. Development and evaluation of deep learning models for multi-modal data analysis.

Introduction:  Multi-modal data analysis is the task of analyzing and understanding data from multiple sources such as text, images, and audio. Deep learning has been used to improve multi-modal data analysis, by training models on large datasets of multi-modal data. This thesis aims to develop and evaluate deep learning models for multi-modal data analysis and analyze their potential to improve performance in comparison to single-modal models.

I hope that this article has provided you with a useful guide for your thesis research in machine learning and deep learning. Remember to conduct a thorough literature review and to include proper citations in your work, as well as to be original in your research to avoid plagiarism. I wish you all the best of luck with your thesis and your research endeavors!

Continue Learning

The depth i: stereo calibration and rectification, using chains and agents for llm application development.

Step-by-step guide to using chains and agents in LangChain

Amazon BedRock — Build Generative AI at Scale

Best practices for ai professional headshots: mastering your visual brand, art generating ai, prompt engineering: how to turn your words into works of art.

• Advisers & Contacts
• Bachelor of Arts & Bachelor of Science in Engineering
• Prerequisites
• Declaring Computer Science for AB Students
• Declaring Computer Science for BSE Students
• Class of '25, '26 & '27 - Departmental Requirements
• Class of 2024 - Departmental Requirements
• COS126 Information
• Important Steps and Deadlines
• Independent Work Seminars
• Guidelines and Useful Information

Undergraduate Research Topics

• AB Junior Research Workshops
• Undergraduate Program FAQ
• How to Enroll
• Requirements
• Certificate Program FAQ
• Interdepartmental Committee
• Minor Program
• Funding for Student Group Activities
• Mailing Lists and Policies
• Study Abroad
• Jobs & Careers
• Admissions Requirements
• Breadth Requirements
• Pre-FPO Checklist
• FPO Checklist
• M.S.E. Track
• M.Eng. Track
• Departmental Internship Policy (for Master's students)
• General Examination
• Fellowship Opportunities
• Travel Reimbursement Policy
• Communication Skills
• Course Schedule
• Course Catalog
• Research Areas
• Interdisciplinary Programs
• Technical Reports
• Computing Facilities
• Researchers
• Technical Staff
• Administrative Staff
• Graduate Students
• Undergraduate Students
• Graduate Alumni
• Climate and Inclusion Committee
• Resources for Undergraduate & Graduate Students
• Outreach Initiatives
• Resources for Faculty & Staff
• Spotlight Stories
• Job Openings
• Undergraduate Program
• Independent Work & Theses

Suggested Undergraduate Research Topics

How to Contact Faculty for IW/Thesis Advising

Send the professor an e-mail. When you write a professor, be clear that you want a meeting regarding a senior thesis or one-on-one IW project, and briefly describe the topic or idea that you want to work on. Check the faculty listing for email addresses.

Parastoo Abtahi, Room 419

Available for single-semester IW and senior thesis advising, 2024-2025

• Research Areas: Human-Computer Interaction (HCI), Augmented Reality (AR), and Spatial Computing
• Input techniques for on-the-go interaction (e.g., eye-gaze, microgestures, voice) with a focus on uncertainty, disambiguation, and privacy.
• Minimal and timely multisensory output (e.g., spatial audio, haptics) that enables users to attend to their physical environment and the people around them, instead of a 2D screen.
• Interaction with intelligent systems (e.g., IoT, robots) situated in physical spaces with a focus on updating users’ mental model despite the complexity and dynamicity of these systems.

Ryan Adams, Room 411

Research areas:

• Machine learning driven design
• Generative models for structured discrete objects
• Approximate inference in probabilistic models
• Accelerating solutions to partial differential equations
• Innovative uses of automatic differentiation
• Modeling and optimizing 3d printing and CNC machining

Andrew Appel, Room 209

Available for Fall 2024 IW advising, only

• Research Areas: Formal methods, programming languages, compilers, computer security.
• Software verification (for which taking COS 326 / COS 510 is helpful preparation)
• Game theory of poker or other games (for which COS 217 / 226 are helpful)
• Computer game-playing programs (for which COS 217 / 226)
•  Risk-limiting audits of elections (for which ORF 245 or other knowledge of probability is useful)

Sanjeev Arora, Room 407

• Theoretical machine learning, deep learning and its analysis, natural language processing. My advisees would typically have taken a course in algorithms (COS423 or COS 521 or equivalent) and a course in machine learning.
• Show that finding approximate solutions to NP-complete problems is also NP-complete (i.e., come up with NP-completeness reductions a la COS 487). 
• Experimental Algorithms: Implementing and Evaluating Algorithms using existing software packages. 
• Studying/designing provable algorithms for machine learning and implementions using packages like scipy and MATLAB, including applications in Natural language processing and deep learning.
• Any topic in theoretical computer science.

David August, Room 221

Not available for IW or thesis advising, 2024-2025

• Research Areas: Computer Architecture, Compilers, Parallelism
• Containment-based approaches to security:  We have designed and tested a simple hardware+software containment mechanism that stops incorrect communication resulting from faults, bugs, or exploits from leaving the system.   Let's explore ways to use containment to solve real problems.  Expect to work with corporate security and technology decision-makers.
• Parallelism: Studies show much more parallelism than is currently realized in compilers and architectures.  Let's find ways to realize this parallelism.
• Any other interesting topic in computer architecture or compilers. 

Mark Braverman, 194 Nassau St., Room 231

• Research Areas: computational complexity, algorithms, applied probability, computability over the real numbers, game theory and mechanism design, information theory.
• Topics in computational and communication complexity.
• Applications of information theory in complexity theory.
• Algorithms for problems under real-life assumptions.
• Game theory, network effects
• Mechanism design (could be on a problem proposed by the student)

Sebastian Caldas, 221 Nassau Street, Room 105

• Research Areas: collaborative learning, machine learning for healthcare. Typically, I will work with students that have taken COS324.
• Methods for collaborative and continual learning.
• Machine learning for healthcare applications.

Bernard Chazelle, 194 Nassau St., Room 301

• Research Areas: Natural Algorithms, Computational Geometry, Sublinear Algorithms. 
• Natural algorithms (flocking, swarming, social networks, etc).
• Sublinear algorithms
• Self-improving algorithms
• Markov data structures

Danqi Chen, Room 412

• My advisees would be expected to have taken a course in machine learning and ideally have taken COS484 or an NLP graduate seminar.
• Representation learning for text and knowledge bases
• Pre-training and transfer learning
• Question answering and reading comprehension
• Information extraction
• Text summarization
• Any other interesting topics related to natural language understanding/generation

Marcel Dall'Agnol, Corwin 034

• Research Areas: Theoretical computer science. (Specifically, quantum computation, sublinear algorithms, complexity theory, interactive proofs and cryptography)
• Research Areas: Machine learning

Jia Deng, Room 423

•  Research Areas: Computer Vision, Machine Learning.
• Object recognition and action recognition
• Deep Learning, autoML, meta-learning
• Geometric reasoning, logical reasoning

Adji Bousso Dieng, Room 406

• Research areas: Vertaix is a research lab at Princeton University led by Professor Adji Bousso Dieng. We work at the intersection of artificial intelligence (AI) and the natural sciences. The models and algorithms we develop are motivated by problems in those domains and contribute to advancing methodological research in AI. We leverage tools in statistical machine learning and deep learning in developing methods for learning with the data, of various modalities, arising from the natural sciences.

Robert Dondero, Corwin Hall, Room 038

• Research Areas:  Software engineering; software engineering education.
• Develop or evaluate tools to facilitate student learning in undergraduate computer science courses at Princeton, and beyond.
• In particular, can code critiquing tools help students learn about software quality?

Zeev Dvir, 194 Nassau St., Room 250

• Research Areas: computational complexity, pseudo-randomness, coding theory and discrete mathematics.
• Independent Research: I have various research problems related to Pseudorandomness, Coding theory, Complexity and Discrete mathematics - all of which require strong mathematical background. A project could also be based on writing a survey paper describing results from a few theory papers revolving around some particular subject.

Benjamin Eysenbach, Room 416

• Research areas: reinforcement learning, machine learning. My advisees would typically have taken COS324.
• Using RL algorithms to applications in science and engineering.
• Emergent behavior of RL algorithms on high-fidelity robotic simulators.
• Studying how architectures and representations can facilitate generalization.

Christiane Fellbaum, 1-S-14 Green

• Research Areas: theoretical and computational linguistics, word sense disambiguation, lexical resource construction, English and multilingual WordNet(s), ontology
• Anything having to do with natural language--come and see me with/for ideas suitable to your background and interests. Some topics students have worked on in the past:
• Developing parsers, part-of-speech taggers, morphological analyzers for underrepresented languages (you don't have to know the language to develop such tools!)
• Quantitative approaches to theoretical linguistics questions
• Extensions and interfaces for WordNet (English and WN in other languages),
• Applications of WordNet(s), including:
• Foreign language tutoring systems,
• Spelling correction software,
• Word-finding/suggestion software for ordinary users and people with memory problems,
• Machine Translation 
• Sentiment and Opinion detection
• Automatic reasoning and inferencing
• Collaboration with professors in the social sciences and humanities ("Digital Humanities")

Adam Finkelstein, Room 424 

• Research Areas: computer graphics, audio.

Robert S. Fish, Corwin Hall, Room 037

• Networking and telecommunications
• Learning, perception, and intelligence, artificial and otherwise;
• Human-computer interaction and computer-supported cooperative work
• Online education, especially in Computer Science Education
• Topics in research and development innovation methodologies including standards, open-source, and entrepreneurship
• Distributed autonomous organizations and related blockchain technologies

Michael Freedman, Room 308 

• Research Areas: Distributed systems, security, networking
• Projects related to streaming data analysis, datacenter systems and networks, untrusted cloud storage and applications. Please see my group website at http://sns.cs.princeton.edu/ for current research projects.

Ruth Fong, Room 032

• Research Areas: computer vision, machine learning, deep learning, interpretability, explainable AI, fairness and bias in AI
• Develop a technique for understanding AI models
• Design a AI model that is interpretable by design
• Build a paradigm for detecting and/or correcting failure points in an AI model
• Analyze an existing AI model and/or dataset to better understand its failure points
• Build a computer vision system for another domain (e.g., medical imaging, satellite data, etc.)
• Develop a software package for explainable AI
• Adapt explainable AI research to a consumer-facing problem

Note: I am happy to advise any project if there's a sufficient overlap in interest and/or expertise; please reach out via email to chat about project ideas.

Tom Griffiths, Room 405

Available for Fall 2024 single-semester IW advising, only

Research areas: computational cognitive science, computational social science, machine learning and artificial intelligence

Note: I am open to projects that apply ideas from computer science to understanding aspects of human cognition in a wide range of areas, from decision-making to cultural evolution and everything in between. For example, we have current projects analyzing chess game data and magic tricks, both of which give us clues about how human minds work. Students who have expertise or access to data related to games, magic, strategic sports like fencing, or other quantifiable domains of human behavior feel free to get in touch.

Aarti Gupta, Room 220

• Research Areas: Formal methods, program analysis, logic decision procedures
• Finding bugs in open source software using automatic verification tools
• Software verification (program analysis, model checking, test generation)
• Decision procedures for logical reasoning (SAT solvers, SMT solvers)

Elad Hazan, Room 409  

• Research interests: machine learning methods and algorithms, efficient methods for mathematical optimization, regret minimization in games, reinforcement learning, control theory and practice
• Machine learning, efficient methods for mathematical optimization, statistical and computational learning theory, regret minimization in games.
• Implementation and algorithm engineering for control, reinforcement learning and robotics
• Implementation and algorithm engineering for time series prediction

Felix Heide, Room 410

• Research Areas: Computational Imaging, Computer Vision, Machine Learning (focus on Optimization and Approximate Inference).
• Optical Neural Networks
• Hardware-in-the-loop Holography
• Zero-shot and Simulation-only Learning
• Object recognition in extreme conditions
• 3D Scene Representations for View Generation and Inverse Problems
• Long-range Imaging in Scattering Media
• Hardware-in-the-loop Illumination and Sensor Optimization
• Inverse Lidar Design
• Phase Retrieval Algorithms
• Proximal Algorithms for Learning and Inference
• Domain-Specific Language for Optics Design

Peter Henderson , 302 Sherrerd Hall

• Research Areas: Machine learning, law, and policy

Kyle Jamieson, Room 306

• Research areas: Wireless and mobile networking; indoor radar and indoor localization; Internet of Things
• See other topics on my independent work  ideas page  (campus IP and CS dept. login req'd)

Alan Kaplan, 221 Nassau Street, Room 105

Research Areas:

• Random apps of kindness - mobile application/technology frameworks used to help individuals or communities; topic areas include, but are not limited to: first response, accessibility, environment, sustainability, social activism, civic computing, tele-health, remote learning, crowdsourcing, etc.
• Tools automating programming language interoperability - Java/C++, React Native/Java, etc.
• Software visualization tools for education
• Connected consumer devices, applications and protocols

Brian Kernighan, Room 311

• Research Areas: application-specific languages, document preparation, user interfaces, software tools, programming methodology
• Application-oriented languages, scripting languages.
• Tools; user interfaces
• Digital humanities

Zachary Kincaid, Room 219

• Research areas: programming languages, program analysis, program verification, automated reasoning
• Independent Research Topics:
• Develop a practical algorithm for an intractable problem (e.g., by developing practical search heuristics, or by reducing to, or by identifying a tractable sub-problem, ...).
• Design a domain-specific programming language, or prototype a new feature for an existing language.
• Any interesting project related to programming languages or logic.

Gillat Kol, Room 316

• Research area: theory

Aleksandra Korolova, 309 Sherrerd Hall

• Research areas: Societal impacts of algorithms and AI; privacy; fair and privacy-preserving machine learning; algorithm auditing.

Advisees typically have taken one or more of COS 226, COS 324, COS 423, COS 424 or COS 445.

Pravesh Kothari, Room 320

• Research areas: Theory

Amit Levy, Room 307

• Research Areas: Operating Systems, Distributed Systems, Embedded Systems, Internet of Things
• Distributed hardware testing infrastructure
• Second factor security tokens
• Low-power wireless network protocol implementation
• USB device driver implementation

Kai Li, Room 321

• Research Areas: Distributed systems; storage systems; content-based search and data analysis of large datasets.
• Fast communication mechanisms for heterogeneous clusters.
• Approximate nearest-neighbor search for high dimensional data.
• Data analysis and prediction of in-patient medical data.
• Optimized implementation of classification algorithms on manycore processors.

Xiaoyan Li, 221 Nassau Street, Room 104

• Research areas: Information retrieval, novelty detection, question answering, AI, machine learning and data analysis.
• Explore new statistical retrieval models for document retrieval and question answering.
• Apply AI in various fields.
• Apply supervised or unsupervised learning in health, education, finance, and social networks, etc.
• Any interesting project related to AI, machine learning, and data analysis.

Lydia Liu, Room 414

• Research Areas: algorithmic decision making, machine learning and society
• Theoretical foundations for algorithmic decision making (e.g. mathematical modeling of data-driven decision processes, societal level dynamics)
• Societal impacts of algorithms and AI through a socio-technical lens (e.g. normative implications of worst case ML metrics, prediction and model arbitrariness)
• Machine learning for social impact domains, especially education (e.g. responsible development and use of LLMs for education equity and access)
• Evaluation of human-AI decision making using statistical methods (e.g. causal inference of long term impact)

Wyatt Lloyd, Room 323

• Research areas: Distributed Systems
• Caching algorithms and implementations
• Storage systems
• Distributed transaction algorithms and implementations

Alex Lombardi , Room 312

• Research Areas: Theory

Margaret Martonosi, Room 208

• Quantum Computing research, particularly related to architecture and compiler issues for QC.
• Computer architectures specialized for modern workloads (e.g., graph analytics, machine learning algorithms, mobile applications
• Investigating security and privacy vulnerabilities in computer systems, particularly IoT devices.
• Other topics in computer architecture or mobile / IoT systems also possible.

Jonathan Mayer, Sherrerd Hall, Room 307 

Available for Spring 2025 single-semester IW, only

• Research areas: Technology law and policy, with emphasis on national security, criminal procedure, consumer privacy, network management, and online speech.
• Assessing the effects of government policies, both in the public and private sectors.
• Collecting new data that relates to government decision making, including surveying current business practices and studying user behavior.
• Developing new tools to improve government processes and offer policy alternatives.

Mae Milano, Room 307

• Local-first / peer-to-peer systems
• Wide-ares storage systems
• Consistency and protocol design
• Type-safe concurrency
• Language design
• Gradual typing
• Domain-specific languages
• Languages for distributed systems

Andrés Monroy-Hernández, Room 405

• Research Areas: Human-Computer Interaction, Social Computing, Public-Interest Technology, Augmented Reality, Urban Computing
• Research interests:developing public-interest socio-technical systems.  We are currently creating alternatives to gig work platforms that are more equitable for all stakeholders. For instance, we are investigating the socio-technical affordances necessary to support a co-op food delivery network owned and managed by workers and restaurants. We are exploring novel system designs that support self-governance, decentralized/federated models, community-centered data ownership, and portable reputation systems.  We have opportunities for students interested in human-centered computing, UI/UX design, full-stack software development, and qualitative/quantitative user research.
• Beyond our core projects, we are open to working on research projects that explore the use of emerging technologies, such as AR, wearables, NFTs, and DAOs, for creative and out-of-the-box applications.

Christopher Moretti, Corwin Hall, Room 036

• Research areas: Distributed systems, high-throughput computing, computer science/engineering education
• Expansion, improvement, and evaluation of open-source distributed computing software.
• Applications of distributed computing for "big science" (e.g. biometrics, data mining, bioinformatics)
• Software and best practices for computer science education and study, especially Princeton's 126/217/226 sequence or MOOCs development
• Sports analytics and/or crowd-sourced computing

Radhika Nagpal, F316 Engineering Quadrangle

• Research areas: control, robotics and dynamical systems

Karthik Narasimhan, Room 422

• Research areas: Natural Language Processing, Reinforcement Learning
• Autonomous agents for text-based games ( https://www.microsoft.com/en-us/research/project/textworld/ )
• Transfer learning/generalization in NLP
• Techniques for generating natural language
• Model-based reinforcement learning

Arvind Narayanan, 308 Sherrerd Hall 

Research Areas: fair machine learning (and AI ethics more broadly), the social impact of algorithmic systems, tech policy

Pedro Paredes, Corwin Hall, Room 041

My primary research work is in Theoretical Computer Science.

 * Research Interest: Spectral Graph theory, Pseudorandomness, Complexity theory, Coding Theory, Quantum Information Theory, Combinatorics.

The IW projects I am interested in advising can be divided into three categories:

 1. Theoretical research

I am open to advise work on research projects in any topic in one of my research areas of interest. A project could also be based on writing a survey given results from a few papers. Students should have a solid background in math (e.g., elementary combinatorics, graph theory, discrete probability, basic algebra/calculus) and theoretical computer science (226 and 240 material, like big-O/Omega/Theta, basic complexity theory, basic fundamental algorithms). Mathematical maturity is a must.

A (non exhaustive) list of topics of projects I'm interested in:   * Explicit constructions of better vertex expanders and/or unique neighbor expanders.   * Construction deterministic or random high dimensional expanders.   * Pseudorandom generators for different problems.   * Topics around the quantum PCP conjecture.   * Topics around quantum error correcting codes and locally testable codes, including constructions, encoding and decoding algorithms.

 2. Theory informed practical implementations of algorithms   Very often the great advances in theoretical research are either not tested in practice or not even feasible to be implemented in practice. Thus, I am interested in any project that consists in trying to make theoretical ideas applicable in practice. This includes coming up with new algorithms that trade some theoretical guarantees for feasible implementation yet trying to retain the soul of the original idea; implementing new algorithms in a suitable programming language; and empirically testing practical implementations and comparing them with benchmarks / theoretical expectations. A project in this area doesn't have to be in my main areas of research, any theoretical result could be suitable for such a project.

Some examples of areas of interest:   * Streaming algorithms.   * Numeric linear algebra.   * Property testing.   * Parallel / Distributed algorithms.   * Online algorithms.    3. Machine learning with a theoretical foundation

I am interested in projects in machine learning that have some mathematical/theoretical, even if most of the project is applied. This includes topics like mathematical optimization, statistical learning, fairness and privacy.

One particular area I have been recently interested in is in the area of rating systems (e.g., Chess elo) and applications of this to experts problems.

Final Note: I am also willing to advise any project with any mathematical/theoretical component, even if it's not the main one; please reach out via email to chat about project ideas.

Iasonas Petras, Corwin Hall, Room 033

• Research Areas: Information Based Complexity, Numerical Analysis, Quantum Computation.
• Prerequisites: Reasonable mathematical maturity. In case of a project related to Quantum Computation a certain familiarity with quantum mechanics is required (related courses: ELE 396/PHY 208).
• Possible research topics include:

1.   Quantum algorithms and circuits:

• i. Design or simulation quantum circuits implementing quantum algorithms.
• ii. Design of quantum algorithms solving/approximating continuous problems (such as Eigenvalue problems for Partial Differential Equations).

2.   Information Based Complexity:

• i. Necessary and sufficient conditions for tractability of Linear and Linear Tensor Product Problems in various settings (for example worst case or average case). 
• ii. Necessary and sufficient conditions for tractability of Linear and Linear Tensor Product Problems under new tractability and error criteria.
• iii. Necessary and sufficient conditions for tractability of Weighted problems.
• iv. Necessary and sufficient conditions for tractability of Weighted Problems under new tractability and error criteria.

3. Topics in Scientific Computation:

• i. Randomness, Pseudorandomness, MC and QMC methods and their applications (Finance, etc)

Yuri Pritykin, 245 Carl Icahn Lab

• Research interests: Computational biology; Cancer immunology; Regulation of gene expression; Functional genomics; Single-cell technologies.
• Potential research projects: Development, implementation, assessment and/or application of algorithms for analysis, integration, interpretation and visualization of multi-dimensional data in molecular biology, particularly single-cell and spatial genomics data.

Benjamin Raphael, Room 309  

• Research interests: Computational biology and bioinformatics; Cancer genomics; Algorithms and machine learning approaches for analysis of large-scale datasets
• Implementation and application of algorithms to infer evolutionary processes in cancer
• Identifying correlations between combinations of genomic mutations in human and cancer genomes
• Design and implementation of algorithms for genome sequencing from new DNA sequencing technologies
• Graph clustering and network anomaly detection, particularly using diffusion processes and methods from spectral graph theory

Vikram Ramaswamy, 035 Corwin Hall

• Research areas: Interpretability of AI systems, Fairness in AI systems, Computer vision.
• Constructing a new method to explain a model / create an interpretable by design model
• Analyzing a current model / dataset to understand bias within the model/dataset
• Proposing new fairness evaluations
• Proposing new methods to train to improve fairness
• Developing synthetic datasets for fairness / interpretability benchmarks
• Understanding robustness of models

Ran Raz, Room 240

• Research Area: Computational Complexity
• Independent Research Topics: Computational Complexity, Information Theory, Quantum Computation, Theoretical Computer Science

Szymon Rusinkiewicz, Room 406

• Research Areas: computer graphics; computer vision; 3D scanning; 3D printing; robotics; documentation and visualization of cultural heritage artifacts
• Research ways of incorporating rotation invariance into computer visiontasks such as feature matching and classification
• Investigate approaches to robust 3D scan matching
• Model and compensate for imperfections in 3D printing
• Given a collection of small mobile robots, apply control policies learned in simulation to the real robots.

Olga Russakovsky, Room 408

• Research Areas: computer vision, machine learning, deep learning, crowdsourcing, fairness&bias in AI
• Design a semantic segmentation deep learning model that can operate in a zero-shot setting (i.e., recognize and segment objects not seen during training)
• Develop a deep learning classifier that is impervious to protected attributes (such as gender or race) that may be erroneously correlated with target classes
• Build a computer vision system for the novel task of inferring what object (or part of an object) a human is referring to when pointing to a single pixel in the image. This includes both collecting an appropriate dataset using crowdsourcing on Amazon Mechanical Turk, creating a new deep learning formulation for this task, and running extensive analysis of both the data and the model

Sebastian Seung, Princeton Neuroscience Institute, Room 153

• Research Areas: computational neuroscience, connectomics, "deep learning" neural networks, social computing, crowdsourcing, citizen science
• Gamification of neuroscience (EyeWire  2.0)
• Semantic segmentation and object detection in brain images from microscopy
• Computational analysis of brain structure and function
• Neural network theories of brain function

Jaswinder Pal Singh, Room 324

• Research Areas: Boundary of technology and business/applications; building and scaling technology companies with special focus at that boundary; parallel computing systems and applications: parallel and distributed applications and their implications for software and architectural design; system software and programming environments for multiprocessors.
• Develop a startup company idea, and build a plan/prototype for it.
• Explore tradeoffs at the boundary of technology/product and business/applications in a chosen area.
• Study and develop methods to infer insights from data in different application areas, from science to search to finance to others. 
• Design and implement a parallel application. Possible areas include graphics, compression, biology, among many others. Analyze performance bottlenecks using existing tools, and compare programming models/languages.
• Design and implement a scalable distributed algorithm.

Mona Singh, Room 420

• Research Areas: computational molecular biology, as well as its interface with machine learning and algorithms.
• Whole and cross-genome methods for predicting protein function and protein-protein interactions.
• Analysis and prediction of biological networks.
• Computational methods for inferring specific aspects of protein structure from protein sequence data.
• Any other interesting project in computational molecular biology.

Robert Tarjan, 194 Nassau St., Room 308

• Research Areas: Data structures; graph algorithms; combinatorial optimization; computational complexity; computational geometry; parallel algorithms.
• Implement one or more data structures or combinatorial algorithms to provide insight into their empirical behavior.
• Design and/or analyze various data structures and combinatorial algorithms.

Olga Troyanskaya, Room 320

• Research Areas: Bioinformatics; analysis of large-scale biological data sets (genomics, gene expression, proteomics, biological networks); algorithms for integration of data from multiple data sources; visualization of biological data; machine learning methods in bioinformatics.
• Implement and evaluate one or more gene expression analysis algorithm.
• Develop algorithms for assessment of performance of genomic analysis methods.
• Develop, implement, and evaluate visualization tools for heterogeneous biological data.

David Walker, Room 211

• Research Areas: Programming languages, type systems, compilers, domain-specific languages, software-defined networking and security
• Independent Research Topics:  Any other interesting project that involves humanitarian hacking, functional programming, domain-specific programming languages, type systems, compilers, software-defined networking, fault tolerance, language-based security, theorem proving, logic or logical frameworks.

Shengyi Wang, Postdoctoral Research Associate, Room 216

Available for Fall 2024 single-semester IW, only

• Independent Research topics: Explore Escher-style tilings using (introductory) group theory and automata theory to produce beautiful pictures.

Kevin Wayne, Corwin Hall, Room 040

• Research Areas: design, analysis, and implementation of algorithms; data structures; combinatorial optimization; graphs and networks.
• Design and implement computer visualizations of algorithms or data structures.
• Develop pedagogical tools or programming assignments for the computer science curriculum at Princeton and beyond.
• Develop assessment infrastructure and assessments for MOOCs.

Matt Weinberg, 194 Nassau St., Room 222

• Research Areas: algorithms, algorithmic game theory, mechanism design, game theoretical problems in {Bitcoin, networking, healthcare}.
• Theoretical questions related to COS 445 topics such as matching theory, voting theory, auction design, etc. 
• Theoretical questions related to incentives in applications like Bitcoin, the Internet, health care, etc. In a little bit more detail: protocols for these systems are often designed assuming that users will follow them. But often, users will actually be strictly happier to deviate from the intended protocol. How should we reason about user behavior in these protocols? How should we design protocols in these settings?

Huacheng Yu, Room 310

• data structures
• streaming algorithms
• design and analyze data structures / streaming algorithms
• prove impossibility results (lower bounds)
• implement and evaluate data structures / streaming algorithms

Ellen Zhong, Room 314

Opportunities outside the department.

We encourage students to look in to doing interdisciplinary computer science research and to work with professors in departments other than computer science.  However, every CS independent work project must have a strong computer science element (even if it has other scientific or artistic elements as well.)  To do a project with an adviser outside of computer science you must have permission of the department.  This can be accomplished by having a second co-adviser within the computer science department or by contacting the independent work supervisor about the project and having he or she sign the independent work proposal form.

Here is a list of professors outside the computer science department who are eager to work with computer science undergraduates.

Maria Apostolaki, Engineering Quadrangle, C330

• Research areas: Computing & Networking, Data & Information Science, Security & Privacy

Branko Glisic, Engineering Quadrangle, Room E330

• Documentation of historic structures
• Cyber physical systems for structural health monitoring
• Developing virtual and augmented reality applications for documenting structures
• Applying machine learning techniques to generate 3D models from 2D plans of buildings
•  Contact : Rebecca Napolitano, rkn2 (@princeton.edu)

Mihir Kshirsagar, Sherrerd Hall, Room 315

Center for Information Technology Policy.

• Consumer protection
• Content regulation
• Competition law
• Economic development
• Surveillance and discrimination

Sharad Malik, Engineering Quadrangle, Room B224

Select a Senior Thesis Adviser for the 2020-21 Academic Year.

• Design of reliable hardware systems
• Verifying complex software and hardware systems

Prateek Mittal, Engineering Quadrangle, Room B236

• Internet security and privacy 
• Social Networks
• Privacy technologies, anonymous communication
• Network Science
• Internet security and privacy: The insecurity of Internet protocols and services threatens the safety of our critical network infrastructure and billions of end users. How can we defend end users as well as our critical network infrastructure from attacks?
• Trustworthy social systems: Online social networks (OSNs) such as Facebook, Google+, and Twitter have revolutionized the way our society communicates. How can we leverage social connections between users to design the next generation of communication systems?
• Privacy Technologies: Privacy on the Internet is eroding rapidly, with businesses and governments mining sensitive user information. How can we protect the privacy of our online communications? The Tor project (https://www.torproject.org/) is a potential application of interest.

Ken Norman,  Psychology Dept, PNI 137

• Research Areas: Memory, the brain and computation 
• Lab:  Princeton Computational Memory Lab

Potential research topics

• Methods for decoding cognitive state information from neuroimaging data (fMRI and EEG) 
• Neural network simulations of learning and memory

Caroline Savage

Office of Sustainability, Phone:(609)258-7513, Email: cs35 (@princeton.edu)

The  Campus as Lab  program supports students using the Princeton campus as a living laboratory to solve sustainability challenges. The Office of Sustainability has created a list of campus as lab research questions, filterable by discipline and topic, on its  website .

An example from Computer Science could include using  TigerEnergy , a platform which provides real-time data on campus energy generation and consumption, to study one of the many energy systems or buildings on campus. Three CS students used TigerEnergy to create a  live energy heatmap of campus .

Other potential projects include:

• Apply game theory to sustainability challenges
• Develop a tool to help visualize interactions between complex campus systems, e.g. energy and water use, transportation and storm water runoff, purchasing and waste, etc.
• How can we learn (in aggregate) about individuals’ waste, energy, transportation, and other behaviors without impinging on privacy?

Janet Vertesi, Sociology Dept, Wallace Hall, Room 122

• Research areas: Sociology of technology; Human-computer interaction; Ubiquitous computing.
• Possible projects: At the intersection of computer science and social science, my students have built mixed reality games, produced artistic and interactive installations, and studied mixed human-robot teams, among other projects.

David Wentzlaff, Engineering Quadrangle, Room 228

Computing, Operating Systems, Sustainable Computing.

• Instrument Princeton's Green (HPCRC) data center
• Investigate power utilization on an processor core implemented in an FPGA
• Dismantle and document all of the components in modern electronics. Invent new ways to build computers that can be recycled easier.
• Other topics in parallel computer architecture or operating systems

Google Custom Search

Wir verwenden Google für unsere Suche. Mit Klick auf „Suche aktivieren“ aktivieren Sie das Suchfeld und akzeptieren die Nutzungsbedingungen.

Hinweise zum Einsatz der Google Suche

• Data Analytics and Machine Learning Group
• TUM School of Computation, Information and Technology
• Technical University of Munich

Open Topics

We offer multiple Bachelor/Master theses, Guided Research projects and IDPs in the area of data mining/machine learning. A  non-exhaustive list of open topics is listed below.

If you are interested in a thesis or a guided research project, please send your CV and transcript of records to Prof. Stephan Günnemann via email and we will arrange a meeting to talk about the potential topics.

Graph Neural Networks for Spatial Transcriptomics

Type:  Master's Thesis

Prerequisites:

• Strong machine learning knowledge
• Proficiency with Python and deep learning frameworks (PyTorch, TensorFlow, JAX)
• Knowledge of graph neural networks (e.g., GCN, MPNN)
• Optional: Knowledge of bioinformatics and genomics

Description:

Spatial transcriptomics is a cutting-edge field at the intersection of genomics and spatial analysis, aiming to understand gene expression patterns within the context of tissue architecture. Our project focuses on leveraging graph neural networks (GNNs) to unlock the full potential of spatial transcriptomic data. Unlike traditional methods, GNNs can effectively capture the intricate spatial relationships between cells, enabling more accurate modeling and interpretation of gene expression dynamics across tissues. We seek motivated students to explore novel GNN architectures tailored for spatial transcriptomics, with a particular emphasis on addressing challenges such as spatial heterogeneity, cell-cell interactions, and spatially varying gene expression patterns.

Contact : Filippo Guerranti , Alessandro Palma

References:

• Cell clustering for spatial transcriptomics data with graph neural network
• Unsupervised spatially embedded deep representation of spatial transcriptomics
• SpaGCN: Integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network
• DeepST: identifying spatial domains in spatial transcriptomics by deep learning
• Deciphering spatial domains from spatially resolved transcriptomics with an adaptive graph attention auto-encoder

GCNG: graph convolutional networks for inferring gene interaction from spatial transcriptomics data

Generative Models for Drug Discovery

Type:  Mater Thesis / Guided Research

• Proficiency with Python and deep learning frameworks (PyTorch or TensorFlow)
• Knowledge of graph neural networks (e.g. GCN, MPNN)
• No formal education in chemistry, physics or biology needed!

Effectively designing molecular geometries is essential to advancing pharmaceutical innovations, a domain which has experienced great attention through the success of generative models. These models promise a more efficient exploration of the vast chemical space and generation of novel compounds with specific properties by leveraging their learned representations, potentially leading to the discovery of molecules with unique properties that would otherwise go undiscovered. Our topics lie at the intersection of generative models like diffusion/flow matching models and graph representation learning, e.g., graph neural networks. The focus of our projects can be model development with an emphasis on downstream tasks ( e.g., diffusion guidance at inference time ) and a better understanding of the limitations of existing models.

Contact :  Johanna Sommer , Leon Hetzel

Equivariant Diffusion for Molecule Generation in 3D

Equivariant Flow Matching with Hybrid Probability Transport for 3D Molecule Generation

Structure-based Drug Design with Equivariant Diffusion Models

Efficient Machine Learning: Pruning, Quantization, Distillation, and More - DAML x Pruna AI

Type: Master's Thesis / Guided Research / Hiwi

• Strong knowledge in machine learning
• Proficiency with Python and deep learning frameworks (TensorFlow or PyTorch)

The efficiency of machine learning algorithms is commonly evaluated by looking at target performance, speed and memory footprint metrics. Reduce the costs associated to these metrics is of primary importance for real-world applications with limited ressources (e.g. embedded systems, real-time predictions). In this project, you will work in collaboration with the DAML research group and the Pruna AI startup on investigating solutions to improve the efficiency of machine leanring models by looking at multiple techniques like pruning, quantization, distillation, and more.

Contact: Bertrand Charpentier

• The Efficiency Misnomer
• A Gradient Flow Framework for Analyzing Network Pruning
• Distilling the Knowledge in a Neural Network
• A Survey of Quantization Methods for Efficient Neural Network Inference

Deep Generative Models

Type:  Master Thesis / Guided Research

• Strong machine learning and probability theory knowledge
• Knowledge of generative models and their basics (e.g., Normalizing Flows, Diffusion Models, VAE)
• Optional: Neural ODEs/SDEs, Optimal Transport, Measure Theory

With recent advances, such as Diffusion Models, Transformers, Normalizing Flows, Flow Matching, etc., the field of generative models has gained significant attention in the machine learning and artificial intelligence research community. However, many problems and questions remain open, and the application to complex data domains such as graphs, time series, point processes, and sets is often non-trivial. We are interested in supervising motivated students to explore and extend the capabilities of state-of-the-art generative models for various data domains.

Contact : Marcel Kollovieh , David Lüdke

• Flow Matching for Generative Modeling
• Auto-Encoding Variational Bayes
• Denoising Diffusion Probabilistic Models 
• Structured Denoising Diffusion Models in Discrete State-Spaces

Active Learning for Multi Agent 3D Object Detection 

Type: Master's Thesis  Industrial partner: BMW 

Prerequisites: 

• Strong knowledge in machine learning 
• Knowledge in Object Detection 
• Excellent programming skills 
• Proficiency with Python and deep learning frameworks (TensorFlow or PyTorch) 

Description: 

In autonomous driving, state-of-the-art deep neural networks are used for perception tasks like for example 3D object detection. To provide promising results, these networks often require a lot of complex annotation data for training. These annotations are often costly and redundant. Active learning is used to select the most informative samples for annotation and cover a dataset with as less annotated data as possible.   

The objective is to explore active learning approaches for 3D object detection using combined uncertainty and diversity based methods.  

Contact: Sebastian Schmidt

References: 

• Exploring Diversity-based Active Learning for 3D Object Detection in Autonomous Driving   
• Efficient Uncertainty Estimation for Semantic Segmentation in Videos   
• KECOR: Kernel Coding Rate Maximization for Active 3D Object Detection
• Towards Open World Active Learning for 3D Object Detection   

Graph Neural Networks

Type:  Master's thesis / Bachelor's thesis / guided research

• Knowledge of graph/network theory

Graph neural networks (GNNs) have recently achieved great successes in a wide variety of applications, such as chemistry, reinforcement learning, knowledge graphs, traffic networks, or computer vision. These models leverage graph data by updating node representations based on messages passed between nodes connected by edges, or by transforming node representation using spectral graph properties. These approaches are very effective, but many theoretical aspects of these models remain unclear and there are many possible extensions to improve GNNs and go beyond the nodes' direct neighbors and simple message aggregation.

Contact: Simon Geisler

• Semi-supervised classification with graph convolutional networks
• Relational inductive biases, deep learning, and graph networks
• Diffusion Improves Graph Learning
• Weisfeiler and leman go neural: Higher-order graph neural networks
• Reliable Graph Neural Networks via Robust Aggregation

Physics-aware Graph Neural Networks

Type:  Master's thesis / guided research

• Proficiency with Python and deep learning frameworks (JAX or PyTorch)
• Knowledge of graph neural networks (e.g. GCN, MPNN, SchNet)
• Optional: Knowledge of machine learning on molecules and quantum chemistry

Deep learning models, especially graph neural networks (GNNs), have recently achieved great successes in predicting quantum mechanical properties of molecules. There is a vast amount of applications for these models, such as finding the best method of chemical synthesis or selecting candidates for drugs, construction materials, batteries, or solar cells. However, GNNs have only been proposed in recent years and there remain many open questions about how to best represent and leverage quantum mechanical properties and methods.

Contact: Nicholas Gao

• Directional Message Passing for Molecular Graphs
• Neural message passing for quantum chemistry
• Learning to Simulate Complex Physics with Graph Network
• Ab initio solution of the many-electron Schrödinger equation with deep neural networks
• Ab-Initio Potential Energy Surfaces by Pairing GNNs with Neural Wave Functions
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds

Robustness Verification for Deep Classifiers

Type: Master's thesis / Guided research

• Strong machine learning knowledge (at least equivalent to IN2064 plus an advanced course on deep learning)
• Strong background in mathematical optimization (preferably combined with Machine Learning setting)
• Proficiency with python and deep learning frameworks (Pytorch or Tensorflow)
• (Preferred) Knowledge of training techniques to obtain classifiers that are robust against small perturbations in data

Description : Recent work shows that deep classifiers suffer under presence of adversarial examples: misclassified points that are very close to the training samples or even visually indistinguishable from them. This undesired behaviour constraints possibilities of deployment in safety critical scenarios for promising classification methods based on neural nets. Therefore, new training methods should be proposed that promote (or preferably ensure) robust behaviour of the classifier around training samples.

Contact: Aleksei Kuvshinov

References (Background):

• Intriguing properties of neural networks
• Explaining and harnessing adversarial examples
• SoK: Certified Robustness for Deep Neural Networks
• Certified Adversarial Robustness via Randomized Smoothing
• Formal guarantees on the robustness of a classifier against adversarial manipulation
• Towards deep learning models resistant to adversarial attacks
• Provable defenses against adversarial examples via the convex outer adversarial polytope
• Certified defenses against adversarial examples
• Lipschitz-margin training: Scalable certification of perturbation invariance for deep neural networks

Uncertainty Estimation in Deep Learning

Type: Master's Thesis / Guided Research

• Strong knowledge in probability theory

Safe prediction is a key feature in many intelligent systems. Classically, Machine Learning models compute output predictions regardless of the underlying uncertainty of the encountered situations. In contrast, aleatoric and epistemic uncertainty bring knowledge about undecidable and uncommon situations. The uncertainty view can be a substantial help to detect and explain unsafe predictions, and therefore make ML systems more robust. The goal of this project is to improve the uncertainty estimation in ML models in various types of task.

Contact: Tom Wollschläger ,   Dominik Fuchsgruber ,   Bertrand Charpentier

• Can You Trust Your Model’s Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift
• Predictive Uncertainty Estimation via Prior Networks
• Posterior Network: Uncertainty Estimation without OOD samples via Density-based Pseudo-Counts
• Evidential Deep Learning to Quantify Classification Uncertainty
• Weight Uncertainty in Neural Networks

Hierarchies in Deep Learning

Type:  Master's Thesis / Guided Research

Multi-scale structures are ubiquitous in real life datasets. As an example, phylogenetic nomenclature naturally reveals a hierarchical classification of species based on their historical evolutions. Learning multi-scale structures can help to exhibit natural and meaningful organizations in the data and also to obtain compact data representation. The goal of this project is to leverage multi-scale structures to improve speed, performances and understanding of Deep Learning models.

Contact: Marcel Kollovieh , Bertrand Charpentier

• Tree Sampling Divergence: An Information-Theoretic Metricfor Hierarchical Graph Clustering
• Hierarchical Graph Representation Learning with Differentiable Pooling
• Gradient-based Hierarchical Clustering
• Gradient-based Hierarchical Clustering using Continuous Representations of Trees in Hyperbolic Space

Thesis Topics

This list includes topics for potential bachelor or master theses, guided research, projects, seminars, and other activities. Search with Ctrl+F for desired keywords, e.g. ‘machine learning’ or others.

PLEASE NOTE: If you are interested in any of these topics, click the respective supervisor link to send a message with a simple CV, grade sheet, and topic ideas (if any). We will answer shortly.

Of course, your own ideas are always welcome!

Image Super-Resolution in both ways

Type of work:.

• Guided Research
• auto-encoder
• deep learning
• single image super-resolution

Description:

The goal of this project is to develop and evaluate a novel dual-decoder architecture for image super-resolution (SR) [1]. This architecture will utilize a single encoder to extract features from an input image, followed by two decoders: one trained to map the features to a low-resolution (LR) output, and the other to map the features to a high-resolution (HR) output. This approach aims to enhance the SR performance by leveraging the complementary learning objectives of both decoders. The goal of the work is to try different architectures and to analyze different loss formulations as well as the feature space learned by the encoder.

• [1] Hitchhiker’s Guide to Super-Resolution: Introduction and Recent Advances

Applying TaylorShift to Transfomer-based Image Super-Resolution Models

• vision transformer

The aim of this project is to integrate the TaylorShift [1] attention mechanism into the SwinIR model to enhance the efficiency and performance of image super-resolution (SR) [2]. By leveraging the linear complexity of TaylorShift, we intend to improve the processing speed and reduce the memory footprint of SwinIR without compromising its high accuracy in generating high-resolution images from low-resolution inputs. Image super-resolution is a crucial task in computer vision that aims to enhance the resolution of images, making them clearer and more detailed. SwinIR (Swin Transformer for Image Restoration) has shown state-of-the-art performance in various image restoration tasks, including super-resolution. However, the quadratic complexity of its attention mechanism can be a bottleneck, especially for high-resolution images. TaylorShift, a novel reformulation of the Taylor softmax function, addresses this issue by reducing the complexity of the attention mechanism from quadratic to linear. This enables efficient processing of long sequences and high-resolution images while maintaining the ability to capture intricate token-to-token interactions.

• [1] TaylorShift: Shifting the Complexity of Self-Attention from Squared to Linear (and Back) using Taylor-Softmax
• [2] Hitchhiker’s Guide to Super-Resolution: Introduction and Recent Advances

Machine Learning-based Surrogate Models for Accelerated Flow Simulations

• Machine Learning
• Microstructure Property Prediction
• Surrogate Modeling

Surrogate modeling involves creating a simplified and computationally efficient machine learning model that approximates the behavior of a complex system, enabling faster predictions and analysis. For complex systems such as fluids, their behavior is governed by partial differential equations. By solving these PDEs, one can predict how a fluid behaves in a specific environment and conditions. The computational time and resources needed to solve a PDE system depend on the size of the fluid domain and the complexity of the PDE. In practical applications where multiple environments and conditions are to be studied, it becomes very expensive to generate many solutions to such PDEs. Here, modern machine learning or deep learning-based surrogate models which offer fast inference times in the online phase are of interest.

In this work, the focus will be on developing surrogate models to replace the flow simulations in fiber-reinforced composite materials governed by the Navier-Stokes equation. Using a conventional PDE solver, a dataset of reference solutions was generated for supervised learning. In this thesis, your tasks will include the conceptualization and implementation of different ML architectures suited for this task, training and evaluation of the models on the available dataset. You will start with simple fully connected architectures and later extend it to 3D convolutional architectures. Also of interest is the infusion of the available domain knowledge into the ML models, known as physics-informed machine learning.

By applying ML to fluid applications, you will learn to acquire the right amount of domain specific knowledge and analyze your results together with domain experts from the field.

If you are interested, please send me an email with your Curriculum Vitae (CV), your Transcript of records and a short statement about your background in related topics.

References:

• Santos, J.E., Xu, D., Jo, H., Landry, C.J., Prodanović, M., Pyrcz, M.J., 2020. PoreFlow-Net: A 3D convolutional neural network to predict fluid flow through porous media. Advances in Water Resources 138, 103539. https://doi.org/10.1016/j.advwatres.2020.103539
• Kashefi, A., Mukerji, T., 2021. Point-cloud deep learning of porous media for permeability prediction. Physics of Fluids 33, 097109. https://doi.org/10.1063/5.0063904

Sherlock Holmes goes AI - Generative comics art of detective scenes and identikits

• Bias in image generation models
• Deep Learning Frameworks
• Frontend visualization
• Speech-To-Text, Text-to-Image Models
• Transformers, Diffusion Models, Hugging Face

Sherlock Holmes is taking the statement of the witness. The witness is describing the appearance of the perpetrator and the forensic setting they still remember. Your task as the AI investigator will be to generate a comic sketch of the scene and phantom images of the accused person based on the spoken statement of the witness. For this you will use state-of-the-art transformers and visualize the output in an application. As AI investigator you will detect, qualify and quantify bias in the images which are produced by different generation models you have chosen.

This work is embedded in the DFKI KI4Pol lab together with the law enforcement agencies. The stories are fictional you will not work on true crime.

Requirements:

• German level B1/2 or equivalent
• Outstanding academic achievements
• Motivational cover letter

Knowledge Graphs für das Immobilienmanagement

• corporate memory
• knowledge graph

Das Management von Immobilien ist komplex und umfasst verschiedenste Informationsquellen und -objekte zur Durchführung der Prozesse. Ein Corporate Memory kann hier unterstützen in der Analyse und Abbildung des Informationsraums um Wissensdienste zu ermöglichen. Aufgabe ist es, eine Ontologie für das Immobilienmanagement zu entwerfen und beispielhaft ein Szenario zu entwickeln. Für die Materialien und Anwendungspartner sind gute Deutschkenntnisse erforderlich.

Fault and Efficiency Prediction in High Performance Computing

• Master Thesis
• event data modelling
• survival modelling
• time series

High use of resources are thought to be an indirect cause of failures in large cluster systems, but little work has systematically investigated the role of high resource usage on system failures, largely due to the lack of a comprehensive resource monitoring tool which resolves resource use by job and node. This project studies log data of the DFKI Kaiserslautern high performance cluster to consider the predictability of adverse events (node failure, GPU freeze), energy usage and identify the most relevant data within. The second supervisor for this work is Joachim Folz.

Data is available via Prometheus -compatible system:

• Node exporter
• DCGM exporter
• Slurm exporter
• Linking Resource Usage Anomalies with System Failures from Cluster Log Data
• Deep Survival Models

Feel free to reach out if the topic sounds interesting or if you have ideas related to this work. We can then brainstorm a specific research question together. Link to my personal website.

Construction & Application of Enterprise Knowledge Graphs in the E-Invoicing Domain

• Guided Research Project
• knowledge graphs
• knowledge services
• linked data
• semantic web

In recent years knowledge graphs received a lot of attention as well in industry as in science. Knowledge graphs consist of entities and relationships between them and allow integrating new knowledge arbitrarily. Famous instances in industry are knowledge graphs by Microsoft, Google, Facebook or IBM. But beyond these ones, knowledge graphs are also adopted in more domain specific scenarios such as in e-Procurement, e-Invoicing and purchase-to-pay processes. The objective in theses and projects is to explore particular aspects of constructing and/or applying knowledge graphs in the domain of purchase-to-pay processes and e-Invoicing.

Anomaly detection in time-series

• explainability

Working on deep neural networks for making the time-series anomaly detection process more robust. An important aspect of this process is explainability of the decision taken by a network.

Time Series Forecasting Using transformer Networks

• time series forecasting
• transformer networks

Transformer networks have emerged as competent architecture for modeling sequences. This research will primarily focus on using transformer networks for forecasting time series (multivariate/ univariate) and may also involve fusing knowledge into the machine learning architecture.

On This Page

Navigation auf uzh.ch

Department of Informatics Artificial Intelligence and Machine Learning Group

Quicklinks und sprachwechsel, main navigation.

This page lists several ideas for Bachelor's and Master's theses. Most of the thesis proposals are based on recent research, exploring branches that most researchers overlook. Some Bachelor's theses proposals can be extended to Master's theses. While each page lists requirements, these are not strict requirements at the beginning of the thesis, but rather skills to acquire during the thesis. For more details, send me an email.  Prof. Dr. Manuel Günther

uzh-wcms-publications.subpageListDialog.labelUnterseiten

• Open Theses
• Active Theses
• Past Theses

Topics for new Theses

Students are most welcome to develop their own topics, and we are happy to supervise them as long as the topic is anywhere close to our fields of expertise -- otherwise we are not able to provide fruitful input. General topics of interest to the AIML group include (but are not limited to):

• Face Recognition: the identification of a person based on an image or a video of their face
• Facial Attribut Classification: the classification of attributes (gender, hair color, face shape, ...) from facial images
• Open-Set Classification: teaching classifiers to discern objects of classes of no interest
• Adversarial Samples: generating small modifications of correctly classified samples that change the output of classifiers, or preventing these kinds of attacks
• Explainable AI: which parts of the inputs are of great importance to the decision of deep learning systems
• Traditional Features: How to make use of Gabor wavelets or other traditional image processing techniques in deep networks
• Medical Image Processing: Applying of the above methods to medical images; I only host topics provided by other insitutitions (ETH, USZ, Balgrist, Idiap, ...)

Topics that we generally do not supervise are in the area of Natural Language Processing, Social Media, Robotics or Reinforcement Learning since we have other experts on these topics ( Prof. Dr. Martin Volk , Prof. Dr. Anikó Hannák , Prof. Dr. Davide Scaramuzza , Prof. Dr. Giorgia Ramponi ) in our department . If you have a topic in mind and believe that we would be a good supervisor for, please send me an email.  Prof. Dr. Manuel Günther

Requirements and Submissions

It is a requirement to use LaTeX for writing the final thesis document. Students should use the AIML Thesis Template (ZIP, 314 KB) .

Generally, all theses need to be defended, including Master theses (mandatory by the Department rules ) and Bachelor theses. Usually, the defense will be scheduled about 3-5 weeks after the submission of the thesis. Deviating from the recommendations, the time for a Master thesis' defense presentation is 30 minutes followed by 15 minutes of questions, while a Bachelor thesis' defense should take 20-30 minutes of presentation and 10-15 minutes of questions.

Additionally, the source code for the thesis needs to be submitted, approximately at the time of the defense. The source code is typically written in Python and use the PyTorch library. It should be self-contained and make use only of public libraries and data (if possible). Source code needs to be documented. This allows fellow students to make use of previously implemented code.

Bachelor and Master Thesis Opportunities

Finding a supervisor.

Our group supervises as many students as possible, but we often do not have the capacity to supervise all interested students.

Preliminaries

Prerequisites for thesis candidates.

Applicants must have significant experience in machine learning, e.g., as acquired through courses offered by our group (passed with a grade of “good” or better) or an equivalent qualification. This often includes a deep conceptual understanding of machine learning and significant programming experience. The necessary skills vary depending on the thesis topic, e.g., purely theoretical theses require more mathematical than programming skills. Candidates must possess the necessary abilities before applying. Cooperation with an external party (like a company or research institute) is usually only possible when the collaborating party does not significantly reduce scientific autonomy (e.g., regarding data access, topic selection, and thesis trajectory).

Application form

Interested students can contact us with the thesis application form as a PDF, a curriculum vitae or resume, and a cover letter. Please provide evidence of relevant skills. Degree theses are often connected to ongoing research in our group. However, it is also possible for students to suggest their own topics/ideas.

Application channels

1. contacting researchers directly.

Candidates can apply to potential supervisors directly using the standardized form above. In your email, please detail your reasons for reaching out to the particular individual.

How to find suitable researchers:

• Filter researchers by interests using this platform .
• Browse our team overview and research profile .
• Contact the researchers from the section “Thesis supervision offerings” below.

Students may send a follow-up email if a researcher does not respond, as messages can sometimes be overlooked amidst numerous emails. Please wait at least two weeks before writing such a follow-up mail.

2. Applying via a mailing list

In addition to targeted applications via channel (1), candidates can send their application to thesis(∂)ml.tu-berlin.de using the standardized form above. This distribution list reaches most researchers in our group. Responses will be provided only if your application aligns well with the needs of at least one researcher in our group.

3. Contacting the Thesis Coordination Team

If previous application attempts via the outlined channels are unsuccessful, and you meet the necessary thesis candidate prerequisites, you are encouraged to contact the Thesis Domain Coordinators. Please include a brief summary of your prior application attempts in your communication.

Thesis Domain Coordinators:

• Quantum Chemistry / ML for Physical Sciences: Stefan Chmiela
• Explainable AI: Grégoire Montavon
• Probabilistic ML: Shinichi Nakajima
• Biomedical Sensing and Neurotechnology: Alexander von Lühmann
• Kernel Methods: Andreas Ziehe
• Computational Neuroscience: Mina Jamshidi Idaji
• Digital Pathology: Julius Hense
• Multimodal Learning: Jannik Wolff
• Anomaly Detection: Alexander Bauer
• NLP and Digital Humanities: Oliver Eberle
• Graph Neural Networks: Thomas Schnake

For topics not covered by the above domains, or for general inquiries and feedback regarding the application process, please direct your correspondence to thesis.coordination(∂)ml.tu-berlin.de. If students do not receive a reply from the Thesis Domain Coordinators or the mailing list, they may send a follow-up email after a waiting period of at least two weeks.

Thesis supervision offerings

Please be aware that we often keep the titles of available theses confidential, as they frequently pertain to our current, yet-to-be-published research. Group members can access our internal database .

High-level overview of inquiring researchers

Publicly advertised thesis topics, next steps after having found a supervisor.

General information (link)

After having found a supervisor, students prepare a thesis proposal that includes

• the research question and its context/motivation,
• related work,
• preliminary methodological and/or experimental results,
• and formalities such as the number of ECTS credits and the writing time specified in the student’s examination regulations.

The supervisor can help the student with writing the proposal. Students may register their thesis with the examination office once Prof. Müller or an authorized senior researcher approves the proposal. We encourage students not to underestimate the time required for writing the proposal. Furthermore, consider that we may require some time to review the proposal. Therefore, it is helpful to apply to our group early.

Writing a Thesis

Finding a thesis topic.

Students who are interested in writing a bachelor’s or master’s thesis should begin thinking about possible topics (cf. hot topics for thesis projects on this page ) or propose their own (cf. introduction to IML ). Good research questions often have their origins in scientific papers around the research topics of the IML lab. Be on the look out for new data sources that might help provide new insights into a special IML research topic.

Your Advisor and Your Committee

In order to write a bachelor’s or master’s thesis you must find an member of the IML lab who is willing to be your thesis advisor. You propose your thesis topic together with your advisor to Prof. Sonntag as the first reviewer in your committee. 

How Long Should it Be? How Long Does it Take?

A bachelor’s thesis is generally 30-60 pages, not including the bibliography. A master’s thesis is generally 60-80 pages, not including the bibliography. However, the length will vary according to the topic and the method of analysis, so the appropriate length will be determined by you, your advisor, and your committee.  Students who write a master’s thesis generally do so over two semesters, bachelor’s one semester.

Procedure and Formal Requirements

• You are a student at Oldenburg University, follow these instructions .
• You are a student at Saarland University, follow these instructions .

You must maintain continuous enrollment Oldenburg University or at Saarland University while working on the bachelor’s or master’s thesis. If you are planning to conduct interviews, surveys or do other research involving human subjects, you must obtain prior approval from DFKI.

Here you can find some theses examples .

Here you can find project group examples .

Hot Topics for Thesis Projects

You will implement novel modern approaches in computer vision such as Transfer Learning, Graph Neural Network, or Semi-Supervised Learning to solve important medical decision problems like Breast cancer detection, Chest-(X-Ray/CT) abnormalities diagnosis, or related medical domains. The target is to achieve state-of-the-art performance and the proposed method could be explainable to end users to improve the system’s reliability.

Nguyen, Duy MH, et al. “An Attention Mechanism using Multiple Knowledge Sources for COVID-19 Detection from CT Images.”,  AAAI 2021, Workshop: Trustworthy AI for Healthcare. 

Soberanis-Mukul, Roger D., Nassir Navab, and Shadi Albarqouni. “An Uncertainty-Driven GCN Refinement Strategy for Organ Segmentation.” arXiv preprint arXiv:2012.03352 (2020).

Contact: Duy Nguyen

In this topic, we will investigate important theoretical machine learning problems that have high impacts on several medical applications. It includes but is not limited to optimization formulation to incorporate efficient user’s feedback to boost the performance of trained models besides available training data (active learning), investigate benefits of transfer learning strategies when dealing with scarce data issues in medical problems, or training algorithms to adapt with highly imbalanced data distribution.

Wilder, Bryan, Eric Horvitz, and Ece Kamar. “Learning to complement humans.” arXiv preprint arXiv:2005.00582 (2020).

De, Abir, et al. “Classification Under Human Assistance.” AAAI (2021).

Yao, Huaxiu, et al. “Hierarchically structured meta-learning.” International Conference on Machine Learning. PMLR, 2019.

Requirements: Programming in Python, ideally experience with processing video and audio data

Project description: The aim is to create an annotated dataset of human-to-human dialogue in Youtube cooking videos*, that can serve as a resource for training ML models to generate conversational explanations of the cooking process. This involves the identification of videos with multiple speakers, speaker diarization (partitioning audio and/or transcript according to speaker identity), identification of conversational interaction between the speakers, and investigating if these interactions qualify as ‘conversational explanations’ of the video content

Contact: Mareike Hartmann

Relevant literature:

Speaker diarization: https://arxiv.org/pdf/2101.09624.pdf Potential videos: http://youcook2.eecs.umich.edu/explore Background on ‘conversational explanations’ from an XAI perspective: https://arxiv.org/pdf/1706.07269.pdf (Sec. 5) Note that in this project, we focus on ‘explaining’ the video content rather than model predictions.

*We focus on the process of cooking as there is some related ongoing work at DFKI, but other instructional scenarios are possible.

Requirements: Programming in Python, Pytorch, basic understanding of Deep Learning, ideally some project work on DL / CV / NLP

Project description: The student will experiment with Image Captioning, more specifically testing existing architectures on different datasets. Then, an error analysis can be conducted, in order to find out how the system can be improved.

Contact: Aliki Anagnostopoulou

• Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
• From Show to Tell: A Survey on Deep Learning-based Image Captioning

Requirements: Programming in Python, Pytorch (or Tensorflow)

Project description: The aim of the project is to investigate how explainable NMT methods are. For example, attention weights from the Transformer architecture can be used as alignments, however it is not straight-forward which weights can be used.

• Jointly Learning to Align and Translate with Transformer Models
• Accurate Word Alignment Induction from Neural Machine Translation

Project description: The aim of the project is to investigate active learning strategies applied to relation extraction from clinical documents when using deep learning models.

Contact: Siting Liang

• BiOnt: Deep Learning using Multiple Biomedical Ontologies for Relation Extraction
• Crowdsourcing Ground Truth for Medical Relation Extraction
• Active Learning for Interactive Relation Extraction in a French Newspaper’s Articles

A central finding of preliminary research reveals that different neural network architectures, when trained on the same data distribution, generate diverse attribution maps for local explanations, supporting the assertion that attribution maps are model-dependent [2]. However, it is also understood that these attribution maps, despite their varying origins, can embody certain common characteristics [1].

Given this premise, the proposition for future research is to delve into the development of a novel algorithm that seeks to create attribution maps universally accepted by all models. These models, despite possessing diverse architectures, are based on the same data distribution. This line of enquiry will pave the way towards generating explanations that are devoid of model-dependency or model-bias, thereby privileging model-invariance.

This research aims to bridge the gap between differing neural network architectures, fostering improved communication, data interpretation, and usability. Ultimately, advancements in this field have the potential to significantly propel the evolution of explainable Artificial Intelligence (AI).

Contact: Md Abdul Kadir

[1] Kadir, M. A., Addluri, G. K., & Sonntag, D. (2023). Harmonizing Feature Attributions Across Deep Learning Architectures: Enhancing Interpretability and Consistency.  arXiv preprint arXiv:2307.02150 .

[2] Gupta, A., Saunshi, N., Yu, D., Lyu, K., & Arora, S. (2022). New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound.  Advances in Neural Information Processing Systems ,  35 , 33120-33133.

Colposcopy, a vital method for the diagnosis of cervical pathology, hinges primarily on the visual cues to detect abnormalities and designate regions for biopsies. The conventional method often includes the use of Acetic acid (5%) for highlighting the cells’ nucleus and hence revealing abnormal or pre-cancerous cells, while green filters aid in visualizing blood vessels supplying these regions. However, vast variations in individual practitioner’s experience and expertise may lead to ununiformed assessments.

This research proposal aims to bridge this gap introducing deep learning algorithms, which have shown unprecedented success in image recognition and classification tasks, into colposcopic examinations [1]. The utilization of these machine learning methodologies could allow automatic detection of cancerous or precancerous regions in colposcopic images or videos, automating and standardizing the evaluation process while offering real-time feedback and suggestions during the examination.

[1] Chandran V, Sumithra MG, Karthick A, George T, Deivakani M, Elakkiya B, Subramaniam U, Manoharan S. Diagnosis of Cervical Cancer based on Ensemble Deep Learning Network using Colposcopy Images. Biomed Res Int. 2021 May 4;2021:5584004. doi: 10.1155/2021/5584004. PMID: 33997017; PMCID: PMC8112909.

• Simple Search
• Advanced Search
• Deposit an Item
• Deposit Instructions
• Instructions for Students

Thesis Files

Repository Staff Only: item control page

• Bibliography
• More Referencing guides Blog Automated transliteration Relevant bibliographies by topics
• Automated transliteration
• Relevant bibliographies by topics
• Referencing guides

Dissertations / Theses on the topic 'Machine Learning (ML)'

Create a spot-on reference in apa, mla, chicago, harvard, and other styles.

Consult the top 50 dissertations / theses for your research on the topic 'Machine Learning (ML).'

Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.

You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.

Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.

Holmberg, Lars. "Human In Command Machine Learning." Licentiate thesis, Malmö universitet, Malmö högskola, Institutionen för datavetenskap och medieteknik (DVMT), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-42576.

Nangalia, V. "ML-EWS - Machine Learning Early Warning System : the application of machine learning to predict in-hospital patient deterioration." Thesis, University College London (University of London), 2017. http://discovery.ucl.ac.uk/1565193/.

John, Meenu Mary. "Design Methods and Processes for ML/DL models." Licentiate thesis, Malmö universitet, Institutionen för datavetenskap och medieteknik (DVMT), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-45026.

Tabell, Johnsson Marco, and Ala Jafar. "Efficiency Comparison Between Curriculum Reinforcement Learning & Reinforcement Learning Using ML-Agents." Thesis, Blekinge Tekniska Högskola, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-20218.

Mattsson, Fredrik, and Anton Gustafsson. "Optimize Ranking System With Machine Learning." Thesis, Högskolan i Halmstad, Akademin för informationsteknologi, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:hh:diva-37431.

Kakadost, Naser, and Charif Ramadan. "Empirisk undersökning av ML strategier vid prediktion av cykelflöden baserad på cykeldata och veckodagar." Thesis, Malmö universitet, Fakulteten för teknik och samhälle (TS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-20168.

Sammaritani, Gloria. "Google BigQuery ML. Analisi comparativa di un nuovo framework per il Machine Learning." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2020.

Gustafsson, Sebastian. "Interpretable serious event forecasting using machine learning and SHAP." Thesis, Uppsala universitet, Institutionen för informationsteknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-444363.

Schoenfeld, Brandon J. "Metalearning by Exploiting Granular Machine Learning Pipeline Metadata." BYU ScholarsArchive, 2020. https://scholarsarchive.byu.edu/etd/8730.

Hellborg, Per. "Optimering av datamängder med Machine learning : En studie om Machine learning och Internet of Things." Thesis, Högskolan i Skövde, Institutionen för informationsteknologi, 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:his:diva-13747.

Nämerforslund, Tim. "Machine Learning Adversaries in Video Games : Using reinforcement learning in the Unity Engine to create compelling enemy characters." Thesis, Mittuniversitetet, Institutionen för informationssystem och –teknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-42746.

Mahfouz, Tarek Said. "Construction legal support for differing site conditions (DSC) through statistical modeling and machine learning (ML)." [Ames, Iowa : Iowa State University], 2009.

REPETTO, MARCO. "Black-box supervised learning and empirical assessment: new perspectives in credit risk modeling." Doctoral thesis, Università degli Studi di Milano-Bicocca, 2023. https://hdl.handle.net/10281/402366.

Björkberg, David. "Comparison of cumulative reward withone, two and three layered artificialneural network in a simple environmentwhen using ml-agents." Thesis, Blekinge Tekniska Högskola, Institutionen för datavetenskap, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-21188.

Ucci, Graziano. "The Interstellar Medium of Galaxies: a Machine Learning Approach." Doctoral thesis, Scuola Normale Superiore, 2019. http://hdl.handle.net/11384/85928.

Gilmore, Eugene M. "Learning Interpretable Decision Tree Classifiers with Human in the Loop Learning and Parallel Coordinates." Thesis, Griffith University, 2022. http://hdl.handle.net/10072/418633.

Sridhar, Sabarish. "SELECTION OF FEATURES FOR ML BASED COMMANDING OF AUTONOMOUS VEHICLES." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-287450.

Lagerkvist, Love. "Neural Novelty — How Machine Learning Does Interactive Generative Literature." Thesis, Malmö universitet, Fakulteten för kultur och samhälle (KS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-21222.

Bhogi, Keerthana. "Two New Applications of Tensors to Machine Learning for Wireless Communications." Thesis, Virginia Tech, 2021. http://hdl.handle.net/10919/104970.

Garg, Anushka. "Comparing Machine Learning Algorithms and Feature Selection Techniques to Predict Undesired Behavior in Business Processesand Study of Auto ML Frameworks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-285559.

Hanski, Jari, and Kaan Baris Biçak. "An Evaluation of the Unity Machine Learning Agents Toolkit in Dense and Sparse Reward Video Game Environments." Thesis, Uppsala universitet, Institutionen för speldesign, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-444982.

Stellmar, Justin. "Predicting the Deformation of 3D Printed ABS Plastic Using Machine Learning Regressions." Youngstown State University / OhioLINK, 2020. http://rave.ohiolink.edu/etdc/view?acc_num=ysu1587462911261523.

Krüger, Franz David, and Mohamad Nabeel. "Hyperparameter Tuning Using Genetic Algorithms : A study of genetic algorithms impact and performance for optimization of ML algorithms." Thesis, Mittuniversitetet, Institutionen för informationssystem och –teknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-42404.

Björkman, Desireé. "Machine Learning Evaluation of Natural Language to Computational Thinking : On the possibilities of coding without syntax." Thesis, Uppsala universitet, Institutionen för informationsteknologi, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-424269.

Lundin, Lowe. "Artificial Intelligence for Data Center Power Consumption Optimisation." Thesis, Uppsala universitet, Avdelningen för systemteknik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-447627.

Wessman, Filip. "Advanced Algorithms for Classification and Anomaly Detection on Log File Data : Comparative study of different Machine Learning Approaches." Thesis, Mittuniversitetet, Institutionen för informationssystem och –teknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-43175.

Narmack, Kirilll. "Dynamic Speed Adaptation for Curves using Machine Learning." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-233545.

Mathias, Berggren, and Sonesson Daniel. "Design Optimization in Gas Turbines using Machine Learning : A study performed for Siemens Energy AB." Thesis, Linköpings universitet, Programvara och system, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-173920.

Giuliani, Luca. "Extending the Moving Targets Method for Injecting Constraints in Machine Learning." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/23885/.

Tarullo, Viviana. "Artificial Neural Networks for classification of EMG data in hand myoelectric control." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/19195/.

Wallner, Vanja. "Mapping medical expressions to MedDRA using Natural Language Processing." Thesis, Uppsala universitet, Institutionen för informationsteknologi, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-426916.

Hellberg, Johan, and Kasper Johansson. "Building Models for Prediction and Forecasting of Service Quality." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-295617.

Nardello, Matteo. "Low-Power Smart Devices for the IoT Revolution." Doctoral thesis, Università degli studi di Trento, 2020. http://hdl.handle.net/11572/274371.

Nardello, Matteo. "Low-Power Smart Devices for the IoT Revolution." Doctoral thesis, Università degli studi di Trento, 2020. http://hdl.handle.net/11572/274371.

Michelini, Mattia. "Barcode detection by neural networks on Android mobile platforms." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2020. http://amslaurea.unibo.it/21080/.

Lundström, Robin. "Machine Learning for Air Flow Characterization : An application of Theory-Guided Data Science for Air Fow characterization in an Industrial Foundry." Thesis, Karlstads universitet, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:kau:diva-72782.

Daneshvar, Saman. "User Modeling in Social Media: Gender and Age Detection." Thesis, Université d'Ottawa / University of Ottawa, 2019. http://hdl.handle.net/10393/39535.

Rosell, Felicia. "Tracking a ball during bounce and roll using recurrent neural networks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-239733.

Hallberg, Jesper. "Searching for the charged Higgs boson in the tau nu analysis using Boosted Decision Trees." Thesis, Uppsala universitet, Högenergifysik, 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-301351.

Klingvall, Emelie. "Artificiell intelligens som ett beslutsstöd inom mammografi : En kvalitativ studie om radiologers perspektiv på icke-tekniska utmaningar." Thesis, Högskolan i Skövde, Institutionen för informationsteknologi, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:his:diva-18768.

Alatram, Ala'a A. M. "A forensic framework for detecting denial-of-service attacks in IoT networks using the MQTT protocol." Thesis, Edith Cowan University, Research Online, Perth, Western Australia, 2022. https://ro.ecu.edu.au/theses/2561.

Forssell, Melker, and Gustav Janér. "Product Matching Using Image Similarity." Thesis, Uppsala universitet, Institutionen för informationsteknologi, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-413481.

Bengtsson, Theodor, and Jonas Hägerlöf. "Stora mängder användardata för produktutveckling : Möjligheter och utmaningar vid integrering av stora mängder användardata i produktutvecklingsprocesser." Thesis, KTH, Integrerad produktutveckling, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-297966.

Nordqvist, My. "Classify part of day and snow on the load of timber stacks : A comparative study between partitional clustering and competitive learning." Thesis, Mittuniversitetet, Institutionen för informationssystem och –teknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-42238.

Mele, Matteo. "Convolutional Neural Networks for the Classification of Olive Oil Geographical Origin." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2020.

Hjerpe, Adam. "Computing Random Forests Variable Importance Measures (VIM) on Mixed Numerical and Categorical Data." Thesis, KTH, Skolan för datavetenskap och kommunikation (CSC), 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-185496.

Ahlm, Kristoffer. "IDENTIFIKATION AV RISKINDIKATORER I FINANSIELL INFORMATION MED HJÄLP AV AI/ML : Ökade möjligheter för myndigheter att förebygga ekonomisk brottslighet." Thesis, Umeå universitet, Institutionen för matematik och matematisk statistik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-184818.

Zanghieri, Marcello. "sEMG-based hand gesture recognition with deep learning." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/18112/.

Benatti, Mattia. "Progettazione e Sviluppo di una Piattaforma Multi-Sorgente per l’Ottimizzazione dei Servizi di Emergenza." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021.

Sibelius, Parmbäck Sebastian. "HMMs and LSTMs for On-line Gesture Recognition on the Stylaero Board : Evaluating and Comparing Two Methods." Thesis, Linköpings universitet, Artificiell intelligens och integrerade datorsystem, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-162237.

The chair typically offers various thesis topics each semester in the areas computational statistics, machine learning, data mining, optimization and statistical software. You are welcome to suggest your own topic as well .

Before you apply for a thesis topic make sure that you fit the following profile:

• Knowledge in machine learning.
• Good R or python skills.

Before you start writing your thesis you must look for a supervisor within the group.

Send an email to the contact person listed in the potential theses topics files with the following information:

• Planned starting date of your thesis.
• Thesis topic (of the list of thesis topics or your own suggestion).
• Previously attended classes on machine learning and programming with R.

Your application will only be processed if it contains all required information.

Potential Thesis Topics

[Potential Thesis Topics] [Student Research Projects] [Current Theses] [Completed Theses]

Below is a list of potential thesis topics. Before you start writing your thesis you must look for a supervisor within the group.

Available thesis topics

Disputation.

The disputation of a thesis lasts about 60-90 minutes and consists of two parts. Only the first part is relevant for the grade and takes 30 minutes (bachelor thesis) and 40 minutes (master thesis). Here, the student is expected to summarize his/her main results of the thesis in a presentation. The supervisor(s) will ask questions regarding the content of the thesis in between. In the second part (after the presentation), the supervisors will give detailed feedback and discuss the thesis with the student. This will take about 30 minutes.

• How do I prepare for the disputation?

You have to prepare a presentation and if there is a bigger time gap between handing in your thesis and the disputation you might want to reread your thesis.

• How many slides should I prepare?

That’s up to you, but you have to respect the time limit. Prepariong more than 20 slides for a Bachelor’s presentation and more than 30 slides for a Master’s is VERY likely a very bad idea.

• Where do I present?

Bernd’s office, in front of the big TV. At least one PhD will be present, maybe more. If you want to present in front of a larger audience in the seminar room or the old library, please book the room yourself and inform us.

• English or German?

We do not care, you can choose.

• What do I have to bring with me?

A document (Prüfungsprotokoll) which you get from “Prüfungsamt” (Frau Maxa or Frau Höfner) for the disputation.Your laptop or a USB stick with the presentation. You can also email Bernd a PDF.

• How does the grading work?

The student will be graded regarding the quality of the thesis, the presentation and the oral discussion of the work. The grade is mainly determined by the written thesis itself, but the grade can improve or drop depending on the presentation and your answers to defense questions.

• What should the presentation cover?

The presentation should cover your thesis, including motivation, introduction, description of new methods and results of your research. Please do NOT explain already existing methods in detail here, put more focus on novel work and the results.

• What kind of questions will be asked after the presentation?

The questions will be directly connected to your thesis and related theory.

Student Research Projects

We are always interested in mentoring interesting student research projects. Please contact us directly with an interesting resarch idea. In the future you will also be able to find research project topics below.

Available projects

Currently we are not offering any student research projects.

For more information please visit the official web page Studentische Forschungsprojekte (Lehre@LMU)

Current Theses (With Working Titles)

Completed theses, completed theses (lmu munich), completed theses (supervised by bernd bischl at tu dortmund).

A list of completed theses and new thesis topics from the Computer Vision Group.

Are you about to start a BSc or MSc thesis? Please read our instructions for preparing and delivering your work.

Below we list possible thesis topics for Bachelor and Master students in the areas of Computer Vision, Machine Learning, Deep Learning and Pattern Recognition. The project descriptions leave plenty of room for your own ideas. If you would like to discuss a topic in detail, please contact the supervisor listed below and Prof. Paolo Favaro to schedule a meeting. Note that for MSc students in Computer Science it is required that the official advisor is a professor in CS.

AI deconvolution of light microscopy images

Level: master.

Background Light microscopy became an indispensable tool in life sciences research. Deconvolution is an important image processing step in improving the quality of microscopy images for removing out-of-focus light, higher resolution, and beter signal to noise ratio. Currently classical deconvolution methods, such as regularisation or blind deconvolution, are implemented in numerous commercial software packages and widely used in research. Recently AI deconvolution algorithms have been introduced and being currently actively developed, as they showed a high application potential.

Aim Adaptation of available AI algorithms for deconvolution of microscopy images. Validation of these methods against state-of-the -art commercially available deconvolution software.

Material and Methods Student will implement and further develop available AI deconvolution methods and acquire test microscopy images of different modalities. Performance of developed AI algorithms will be validated against available commercial deconvolution software.

• Al algorithm development and implementation: 50%.
• Data acquisition: 10%.
• Comparison of performance: 40 %.

Requirements

• Interest in imaging.
• Solid knowledge of AI.
• Good programming skills.

Supervisors Paolo Favaro, Guillaume Witz, Yury Belyaev.

Institutes Computer Vison Group, Digital Science Lab, Microscopy imaging Center.

Contact Yury Belyaev, Microscopy imaging Center, [email protected] , + 41 78 899 0110.

Instance segmentation of cryo-ET images

Level: bachelor/master.

In the 1600s, a pioneering Dutch scientist named Antonie van Leeuwenhoek embarked on a remarkable journey that would forever transform our understanding of the natural world. Armed with a simple yet ingenious invention, the light microscope, he delved into uncharted territory, peering through its lens to reveal the hidden wonders of microscopic structures. Fast forward to today, where cryo-electron tomography (cryo-ET) has emerged as a groundbreaking technique, allowing researchers to study proteins within their natural cellular environments. Proteins, functioning as vital nano-machines, play crucial roles in life and understanding their localization and interactions is key to both basic research and disease comprehension. However, cryo-ET images pose challenges due to inherent noise and a scarcity of annotated data for training deep learning models.

Credit: S. Albert et al./PNAS (CC BY 4.0)

To address these challenges, this project aims to develop a self-supervised pipeline utilizing diffusion models for instance segmentation in cryo-ET images. By leveraging the power of diffusion models, which iteratively diffuse information to capture underlying patterns, the pipeline aims to refine and accurately segment cryo-ET images. Self-supervised learning, which relies on unlabeled data, reduces the dependence on extensive manual annotations. Successful implementation of this pipeline could revolutionize the field of structural biology, facilitating the analysis of protein distribution and organization within cellular contexts. Moreover, it has the potential to alleviate the limitations posed by limited annotated data, enabling more efficient extraction of valuable information from cryo-ET images and advancing biomedical applications by enhancing our understanding of protein behavior.

Methods The segmentation pipeline for cryo-electron tomography (cryo-ET) images consists of two stages: training a diffusion model for image generation and training an instance segmentation U-Net using synthetic and real segmentation masks.

    1. Diffusion Model Training:         a. Data Collection: Collect and curate cryo-ET image datasets from the EMPIAR             database (https://www.ebi.ac.uk/empiar/).         b. Architecture Design: Select an appropriate architecture for the diffusion model.         c. Model Evaluation: Cryo-ET experts will help assess image quality and fidelity             through visual inspection and quantitative measures     2. Building the Segmentation dataset:         a. Synthetic and real mask generation: Use the trained diffusion model to generate             synthetic cryo-ET images. The diffusion process will be seeded from either a real             or a synthetic segmentation mask. This will yield to pairs of cryo-ET images and             segmentation masks.     3. Instance Segmentation U-Net Training:         a. Architecture Design: Choose an appropriate instance segmentation U-Net             architecture.         b. Model Evaluation: Evaluate the trained U-Net using precision, recall, and F1             score metrics.

By combining the diffusion model for cryo-ET image generation and the instance segmentation U-Net, this pipeline provides an efficient and accurate approach to segment structures in cryo-ET images, facilitating further analysis and interpretation.

References     1. Kwon, Diana. "The secret lives of cells-as never seen before." Nature 598.7882 (2021):         558-560.     2. Moebel, Emmanuel, et al. "Deep learning improves macromolecule identification in 3D         cellular cryo-electron tomograms." Nature methods 18.11 (2021): 1386-1394.     3. Rice, Gavin, et al. "TomoTwin: generalized 3D localization of macromolecules in         cryo-electron tomograms with structural data mining." Nature Methods (2023): 1-10.

Contacts Prof. Thomas Lemmin Institute of Biochemistry and Molecular Medicine Bühlstrasse 28, 3012 Bern ( [email protected] )

Prof. Paolo Favaro Institute of Computer Science Neubrückstrasse 10 3012 Bern ( [email protected] )

Adding and removing multiple sclerosis lesions with to imaging with diffusion networks

Background multiple sclerosis lesions are the result of demyelination: they appear as dark spots on t1 weighted mri imaging and as bright spots on flair mri imaging.  image analysis for ms patients requires both the accurate detection of new and enhancing lesions, and the assessment of  atrophy via local thickness and/or volume changes in the cortex.  detection of new and growing lesions is possible using deep learning, but made difficult by the relative lack of training data: meanwhile cortical morphometry can be affected by the presence of lesions, meaning that removing lesions prior to morphometry may be more robust.  existing ‘lesion filling’ methods are rather crude, yielding unrealistic-appearing brains where the borders of the removed lesions are clearly visible., aim: denoising diffusion networks are the current gold standard in mri image generation [1]: we aim to leverage this technology to remove and add lesions to existing mri images.  this will allow us to create realistic synthetic mri images for training and validating ms lesion segmentation algorithms, and for investigating the sensitivity of morphometry software to the presence of ms lesions at a variety of lesion load levels., materials and methods: a large, annotated, heterogeneous dataset of mri data from ms patients, as well as images of healthy controls without white matter lesions, will be available for developing the method.  the student will work in a research group with a long track record in applying deep learning methods to neuroimaging data, as well as experience training denoising diffusion networks..

Nature of the Thesis:

Literature review: 10%

Replication of Blob Loss paper: 10%

Implementation of the sliding window metrics:10%

Training on MS lesion segmentation task: 30%

Extension to other datasets: 20%

Results analysis: 20%

Fig. Results of an existing lesion filling algorithm, showing inadequate performance

Requirements:

Interest/Experience with image processing

Python programming knowledge (Pytorch bonus)

Interest in neuroimaging

Supervisor(s):

PD. Dr. Richard McKinley

Institutes: Diagnostic and Interventional Neuroradiology

Center for Artificial Intelligence in Medicine (CAIM), University of Bern

References: [1] Brain Imaging Generation with Latent Diffusion Models , Pinaya et al, Accepted in the Deep Generative Models workshop @ MICCAI 2022 , https://arxiv.org/abs/2209.07162

Contact : PD Dr Richard McKinley, Support Centre for Advanced Neuroimaging ( [email protected] )

Improving metrics and loss functions for targets with imbalanced size: sliding window Dice coefficient and loss.

Background The Dice coefficient is the most commonly used metric for segmentation quality in medical imaging, and a differentiable version of the coefficient is often used as a loss function, in particular for small target classes such as multiple sclerosis lesions.  Dice coefficient has the benefit that it is applicable in instances where the target class is in the minority (for example, in case of segmenting small lesions).  However, if lesion sizes are mixed, the loss and metric is biased towards performance on large lesions, leading smaller lesions to be missed and harming overall lesion detection.  A recently proposed loss function (blob loss[1]) aims to combat this by treating each connected component of a lesion mask separately, and claims improvements over Dice loss on lesion detection scores in a variety of tasks.

Aim: The aim of this thesisis twofold.  First, to benchmark blob loss against a simple, potentially superior loss for instance detection: sliding window Dice loss, in which the Dice loss is calculated over a sliding window across the area/volume of the medical image.  Second, we will investigate whether a sliding window Dice coefficient is better corellated with lesion-wise detection metrics than Dice coefficient and may serve as an alternative metric capturing both global and instance-wise detection.

Materials and Methods: A large, annotated, heterogeneous dataset of MRI data from MS patients will be available for benchmarking the method, as well as our existing codebases for MS lesion segmentation.  Extension of the method to other diseases and datasets (such as covered in the blob loss paper) will make the method more plausible for publication.  The student will work alongside clinicians and engineers carrying out research in multiple sclerosis lesion segmentation, in particular in the context of our running project supported by the CAIM grant.

Fig. An  annotated MS lesion case, showing the variety of lesion sizes

References: [1] blob loss: instance imbalance aware loss functions for semantic segmentation, Kofler et al, https://arxiv.org/abs/2205.08209

Idempotent and partial skull-stripping in multispectral MRI imaging

Background Skull stripping (or brain extraction) refers to the masking of non-brain tissue from structural MRI imaging.  Since 3D MRI sequences allow reconstruction of facial features, many data providers supply data only after skull-stripping, making this a vital tool in data sharing.  Furthermore, skull-stripping is an important pre-processing step in many neuroimaging pipelines, even in the deep-learning era: while many methods could now operate on data with skull present, they have been trained only on skull-stripped data and therefore produce spurious results on data with the skull present.

High-quality skull-stripping algorithms based on deep learning are now widely available: the most prominent example is HD-BET [1].  A major downside of HD-BET is its behaviour on datasets to which skull-stripping has already been applied: in this case the algorithm falsely identifies brain tissue as skull and masks it.  A skull-stripping algorithm F not exhibiting this behaviour would  be idempotent: F(F(x)) = F(x) for any image x.  Furthermore, legacy datasets from before the availability of high-quality skull-stripping algorithms may still contain images which have been inadequately skull-stripped: currently the only solution to improve the skull-stripping on this data is to go back to the original datasource or to manually correct the skull-stripping, which is time-consuming and prone to error. 

Aim: In this project, the student will develop an idempotent skull-stripping network which can also handle partially skull-stripped inputs.  In the best case, the network will operate well on a large subset of the data we work with (e.g. structural MRI, diffusion-weighted MRI, Perfusion-weighted MRI,  susceptibility-weighted MRI, at a variety of field strengths) to maximize the future applicability of the network across the teams in our group.

Materials and Methods: Multiple datasets, both publicly available and internal (encompassing thousands of 3D volumes) will be available. Silver standard reference data for standard sequences at 1.5T and 3T can be generated using existing tools such as HD-BET: for other sequences and field strengths semi-supervised learning or methods improving robustness to domain shift may be employed.  Robustness to partial skull-stripping may be induced by a combination of learning theory and model-based approaches.

Dataset curation: 10%

Idempotent skull-stripping model building: 30%

Modelling of partial skull-stripping:10%

Extension of model to handle partial skull: 30%

Results analysis: 10%

Fig. An example of failed skull-stripping requiring manual correction

References: [1] Isensee, F, Schell, M, Pflueger, I, et al. Automated brain extraction of multisequence MRI using artificial neural networks. Hum Brain Mapp . 2019; 40: 4952– 4964. https://doi.org/10.1002/hbm.24750

Automated leaf detection and leaf area estimation (for Arabidopsis thaliana)

Correlating plant phenotypes such as leaf area or number of leaves to the genotype (i.e. changes in DNA) is a common goal for plant breeders and molecular biologists. Such data can not only help to understand fundamental processes in nature, but also can help to improve ecotypes, e.g., to perform better under climate change, or reduce fertiliser input. However, collecting data for many plants is very time consuming and automated data acquisition is necessary.

The project aims at building a machine learning model to automatically detect plants in top-view images (see examples below), segment their leaves (see Fig C) and to estimate the leaf area. This information will then be used to determine the leaf area of different Arabidopsis ecotypes. The project will be carried out in collaboration with researchers of the Institute of Plant Sciences at the University of Bern. It will also involve the design and creation of a dataset of plant top-views with the corresponding annotation (provided by experts at the Institute of Plant Sciences).

Contact: Prof. Dr. Paolo Favaro ( [email protected] )

Master Projects at the ARTORG Center

The Gerontechnology and Rehabilitation group at the ARTORG Center for Biomedical Engineering is offering multiple MSc thesis projects to students, which are interested in working with real patient data, artificial intelligence and machine learning algorithms. The goal of these projects is to transfer the findings to the clinic in order to solve today’s healthcare problems and thus to improve the quality of life of patients. Assessment of Digital Biomarkers at Home by Radar.  [PDF] Comparison of Radar, Seismograph and Ballistocardiography and to Monitor Sleep at Home.   [PDF] Sentimental Analysis in Speech.  [PDF] Contact: Dr. Stephan Gerber ( [email protected] )

Internship in Computational Imaging at Prophesee

A 6 month intership at Prophesee, Grenoble is offered to a talented Master Student.

The topic of the internship is working on burst imaging following the work of Sam Hasinoff , and exploring ways to improve it using event-based vision.

A compensation to cover the expenses of living in Grenoble is offered. Only students that have legal rights to work in France can apply.

Anyone interested can send an email with the CV to Daniele Perrone ( [email protected] ).

Using machine learning applied to wearables to predict mental health

This Master’s project lies at the intersection of psychiatry and computer science and aims to use machine learning techniques to improve health. Using sensors to detect sleep and waking behavior has as of yet unexplored potential to reveal insights into health.  In this study, we make use of a watch-like device, called an actigraph, which tracks motion to quantify sleep behavior and waking activity. Participants in the study consist of healthy and depressed adolescents and wear actigraphs for a year during which time we query their mental health status monthly using online questionnaires.  For this masters thesis we aim to make use of machine learning methods to predict mental health based on the data from the actigraph. The ability to predict mental health crises based on sleep and wake behavior would provide an opportunity for intervention, significantly impacting the lives of patients and their families. This Masters thesis is a collaboration between Professor Paolo Favaro at the Institute of Computer Science ( [email protected] ) and Dr Leila Tarokh at the Universitäre Psychiatrische Dienste (UPD) ( [email protected] ).  We are looking for a highly motivated individual interested in bridging disciplines. 

Bachelor or Master Projects at the ARTORG Center

The Gerontechnology and Rehabilitation group at the ARTORG Center for Biomedical Engineering is offering multiple BSc- and MSc thesis projects to students, which are interested in working with real patient data, artificial intelligence and machine learning algorithms. The goal of these projects is to transfer the findings to the clinic in order to solve today’s healthcare problems and thus to improve the quality of life of patients. Machine Learning Based Gait-Parameter Extraction by Using Simple Rangefinder Technology.  [PDF] Detection of Motion in Video Recordings   [PDF] Home-Monitoring of Elderly by Radar  [PDF] Gait feature detection in Parkinson's Disease  [PDF] Development of an arthroscopic training device using virtual reality  [PDF] Contact: Dr. Stephan Gerber ( [email protected] ), Michael Single ( [email protected]. ch )

Dynamic Transformer

Level: bachelor.

Visual Transformers have obtained state of the art classification accuracies [ViT, DeiT, T2T, BoTNet]. Mixture of experts could be used to increase the capacity of a neural network by learning instance dependent execution pathways in a network [MoE]. In this research project we aim to push the transformers to their limit and combine their dynamic attention with MoEs, compared to Switch Transformer [Switch], we will use a much more efficient formulation of mixing [CondConv, DynamicConv] and we will use this idea in the attention part of the transformer, not the fully connected layer.

• Input dependent attention kernel generation for better transformer layers.

Publication Opportunity: Dynamic Neural Networks Meets Computer Vision (a CVPR 2021 Workshop)

Extensions:

• The same idea could be extended to other ViT/Transformer based models [DETR, SETR, LSTR, TrackFormer, BERT]

Related Papers:

• Visual Transformers: Token-based Image Representation and Processing for Computer Vision [ViT]
• DeiT: Data-efficient Image Transformers [DeiT]
• Bottleneck Transformers for Visual Recognition [BoTNet]
• Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet [T2TViT]
• Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer [MoE]
• Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity [Switch]
• CondConv: Conditionally Parameterized Convolutions for Efficient Inference [CondConv]
• Dynamic Convolution: Attention over Convolution Kernels [DynamicConv]
• End-to-End Object Detection with Transformers [DETR]
• Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers [SETR]
• End-to-end Lane Shape Prediction with Transformers [LSTR]
• TrackFormer: Multi-Object Tracking with Transformers [TrackFormer]
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding [BERT]

Contact: Sepehr Sameni

Visual Transformers have obtained state of the art classification accuracies for 2d images[ViT, DeiT, T2T, BoTNet]. In this project, we aim to extend the same ideas to 3d data (videos), which requires a more efficient attention mechanism [Performer, Axial, Linformer]. In order to accelerate the training process, we could use [Multigrid] technique.

• Better video understanding by attention blocks.

Publication Opportunity: LOVEU (a CVPR workshop) , Holistic Video Understanding (a CVPR workshop) , ActivityNet (a CVPR workshop)

• Rethinking Attention with Performers [Performer]
• Axial Attention in Multidimensional Transformers [Axial]
• Linformer: Self-Attention with Linear Complexity [Linformer]
• A Multigrid Method for Efficiently Training Video Models [Multigrid]

GIRAFFE is a newly introduced GAN that can generate scenes via composition with minimal supervision [GIRAFFE]. Generative methods can implicitly learn interpretable representation as can be seen in GAN image interpretations [GANSpace, GanLatentDiscovery]. Decoding GIRAFFE could give us per-object interpretable representations that could be used for scene manipulation, data augmentation, scene understanding, semantic segmentation, pose estimation [iNeRF], and more. 

In order to invert a GIRAFFE model, we will first train the generative model on Clevr and CompCars datasets, then we add a decoder to the pipeline and train this autoencoder. We can make the task easier by knowing the number of objects in the scene and/or knowing their positions. 

Goals:  

Scene Manipulation and Decomposition by Inverting the GIRAFFE 

Publication Opportunity:  DynaVis 2021 (a CVPR workshop on Dynamic Scene Reconstruction)  

Related Papers: 

• GIRAFFE: Representing Scenes as Compositional Generative Neural Feature Fields [GIRAFFE] 
• Neural Scene Graphs for Dynamic Scenes 
• pixelNeRF: Neural Radiance Fields from One or Few Images [pixelNeRF] 
• NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis [NeRF] 
• Neural Volume Rendering: NeRF And Beyond 
• GANSpace: Discovering Interpretable GAN Controls [GANSpace] 
• Unsupervised Discovery of Interpretable Directions in the GAN Latent Space [GanLatentDiscovery] 
• Inverting Neural Radiance Fields for Pose Estimation [iNeRF] 

Quantized ViT

Visual Transformers have obtained state of the art classification accuracies [ViT, CLIP, DeiT], but the best ViT models are extremely compute heavy and running them even only for inference (not doing backpropagation) is expensive. Running transformers cheaply by quantization is not a new problem and it has been tackled before for BERT [BERT] in NLP [Q-BERT, Q8BERT, TernaryBERT, BinaryBERT]. In this project we will be trying to quantize pretrained ViT models. 

Quantizing ViT models for faster inference and smaller models without losing accuracy 

Publication Opportunity:  Binary Networks for Computer Vision 2021 (a CVPR workshop)  

Extensions:  

• Having a fast pipeline for image inference with ViT will allow us to dig deep into the attention of ViT and analyze it, we might be able to prune some attention heads or replace them with static patterns (like local convolution or dilated patterns), We might be even able to replace the transformer with performer and increase the throughput even more [Performer]. 
• The same idea could be extended to other ViT based models [DETR, SETR, LSTR, TrackFormer, CPTR, BoTNet, T2TViT] 
• Learning Transferable Visual Models From Natural Language Supervision [CLIP] 
• Visual Transformers: Token-based Image Representation and Processing for Computer Vision [ViT] 
• DeiT: Data-efficient Image Transformers [DeiT] 
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding [BERT] 
• Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT [Q-BERT] 
• Q8BERT: Quantized 8Bit BERT [Q8BERT] 
• TernaryBERT: Distillation-aware Ultra-low Bit BERT [TernaryBERT] 
• BinaryBERT: Pushing the Limit of BERT Quantization [BinaryBERT] 
• Rethinking Attention with Performers [Performer] 
• End-to-End Object Detection with Transformers [DETR] 
• Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers [SETR] 
• End-to-end Lane Shape Prediction with Transformers [LSTR] 
• TrackFormer: Multi-Object Tracking with Transformers [TrackFormer] 
• CPTR: Full Transformer Network for Image Captioning [CPTR] 
• Bottleneck Transformers for Visual Recognition [BoTNet] 
• Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet [T2TViT] 

Multimodal Contrastive Learning

Recently contrastive learning has gained a lot of attention for self-supervised image representation learning [SimCLR, MoCo]. Contrastive learning could be extended to multimodal data, like videos (images and audio) [CMC, CoCLR]. Most contrastive methods require large batch sizes (or large memory pools) which makes them expensive for training. In this project we are going to use non batch size dependent contrastive methods [SwAV, BYOL, SimSiam] to train multimodal representation extractors. 

Our main goal is to compare the proposed method with the CMC baseline, so we will be working with STL10, ImageNet, UCF101, HMDB51, and NYU Depth-V2 datasets. 

Inspired by the recent works on smaller datasets [ConVIRT, CPD], to accelerate the training speed, we could start with two pretrained single-modal models and finetune them with the proposed method.  

• Extending SwAV to multimodal datasets 
• Grasping a better understanding of the BYOL 

Publication Opportunity:  MULA 2021 (a CVPR workshop on Multimodal Learning and Applications)  

• Most knowledge distillation methods for contrastive learners also use large batch sizes (or memory pools) [CRD, SEED], the proposed method could be extended for knowledge distillation. 
• One could easily extend this idea to multiview learning, for example one could have two different networks working on the same input and train them with contrastive learning, this may lead to better models [DeiT] by cross-model inductive biases communications. 
• Self-supervised Co-training for Video Representation Learning [CoCLR] 
• Learning Spatiotemporal Features via Video and Text Pair Discrimination [CPD] 
• Audio-Visual Instance Discrimination with Cross-Modal Agreement [AVID-CMA] 
• Self-Supervised Learning by Cross-Modal Audio-Video Clustering [XDC] 
• Contrastive Multiview Coding [CPC] 
• Contrastive Learning of Medical Visual Representations from Paired Images and Text [ConVIRT] 
• A Simple Framework for Contrastive Learning of Visual Representations [SimCLR] 
• Momentum Contrast for Unsupervised Visual Representation Learning [MoCo] 
• Bootstrap your own latent: A new approach to self-supervised Learning [BYOL] 
• Exploring Simple Siamese Representation Learning [SimSiam] 
• Unsupervised Learning of Visual Features by Contrasting Cluster Assignments [SwAV] 
• Contrastive Representation Distillation [CRD] 
• SEED: Self-supervised Distillation For Visual Representation [SEED] 

Robustness of Neural Networks

Neural Networks have been found to achieve surprising performance in several tasks such as classification, detection and segmentation. However, they are also very sensitive to small (controlled) changes to the input. It has been shown that some changes to an image that are not visible to the naked eye may lead the network to output an incorrect label. This thesis will focus on studying recent progress in this area and aim to build a procedure for a trained network to self-assess its reliability in classification or one of the popular computer vision tasks.

Contact: Paolo Favaro

Masters projects at sitem center

The Personalised Medicine Research Group at the sitem Center for Translational Medicine and Biomedical Entrepreneurship is offering multiple MSc thesis projects to the biomed eng MSc students that may also be of interest to the computer science students. Automated quantification of cartilage quality for hip treatment decision support.  PDF Automated quantification of massive rotator cuff tears from MRI. PDF Deep learning-based segmentation and fat fraction analysis of the shoulder muscles using quantitative MRI. PDF Unsupervised Domain Adaption for Cross-Modality Hip Joint Segmentation. PDF Contact:  Dr. Kate Gerber

Internships/Master thesis @ Chronocam

3-6 months internships on event-based computer vision. Chronocam is a rapidly growing startup developing event-based technology, with more than 15 PhDs working on problems like tracking, detection, classification, SLAM, etc. Event-based computer vision has the potential to solve many long-standing problems in traditional computer vision, and this is a super exciting time as this potential is becoming more and more tangible in many real-world applications. For next year we are looking for motivated Master and PhD students with good software engineering skills (C++ and/or python), and preferable good computer vision and deep learning background. PhD internships will be more research focused and possibly lead to a publication.  For each intern we offer a compensation to cover the expenses of living in Paris.  List of some of the topics we want to explore:

• Photo-realistic image synthesis and super-resolution from event-based data (PhD)
• Self-supervised representation learning (PhD)
• End-to-end Feature Learning for Event-based Data
• Bio-inspired Filtering using Spiking Networks
• On-the fly Compression of Event-based Streams for Low-Power IoT Cameras
• Tracking of Multiple Objects with a Dual-Frequency Tracker
• Event-based Autofocus
• Stabilizing an Event-based Stream using an IMU
• Crowd Monitoring for Low-power IoT Cameras
• Road Extraction from an Event-based Camera Mounted in a Car for Autonomous Driving
• Sign detection from an Event-based Camera Mounted in a Car for Autonomous Driving
• High-frequency Eye Tracking

Email with attached CV to Daniele Perrone at  [email protected] .

Contact: Daniele Perrone

Object Detection in 3D Point Clouds

Today we have many 3D scanning techniques that allow us to capture the shape and appearance of objects. It is easier than ever to scan real 3D objects and transform them into a digital model for further processing, such as modeling, rendering or animation. However, the output of a 3D scanner is often a raw point cloud with little to no annotations. The unstructured nature of the point cloud representation makes it difficult for processing, e.g. surface reconstruction. One application is the detection and segmentation of an object of interest.  In this project, the student is challenged to design a system that takes a point cloud (a 3D scan) as input and outputs the names of objects contained in the scan. This output can then be used to eliminate outliers or points that belong to the background. The approach involves collecting a large dataset of 3D scans and training a neural network on it.

Contact: Adrian Wälchli

Shape Reconstruction from a Single RGB Image or Depth Map

A photograph accurately captures the world in a moment of time and from a specific perspective. Since it is a projection of the 3D space to a 2D image plane, the depth information is lost. Is it possible to restore it, given only a single photograph? In general, the answer is no. This problem is ill-posed, meaning that many different plausible depth maps exist, and there is no way of telling which one is the correct one.  However, if we cover one of our eyes, we are still able to recognize objects and estimate how far away they are. This motivates the exploration of an approach where prior knowledge can be leveraged to reduce the ill-posedness of the problem. Such a prior could be learned by a deep neural network, trained with many images and depth maps.

CNN Based Deblurring on Mobile

Deblurring finds many applications in our everyday life. It is particularly useful when taking pictures on handheld devices (e.g. smartphones) where camera shake can degrade important details. Therefore, it is desired to have a good deblurring algorithm implemented directly in the device.  In this project, the student will implement and optimize a state-of-the-art deblurring method based on a deep neural network for deployment on mobile phones (Android).  The goal is to reduce the number of network weights in order to reduce the memory footprint while preserving the quality of the deblurred images. The result will be a camera app that automatically deblurs the pictures, giving the user a choice of keeping the original or the deblurred image.

Depth from Blur

If an object in front of the camera or the camera itself moves while the aperture is open, the region of motion becomes blurred because the incoming light is accumulated in different positions across the sensor. If there is camera motion, there is also parallax. Thus, a motion blurred image contains depth information.  In this project, the student will tackle the problem of recovering a depth-map from a motion-blurred image. This includes the collection of a large dataset of blurred- and sharp images or videos using a pair or triplet of GoPro action cameras. Two cameras will be used in stereo to estimate the depth map, and the third captures the blurred frames. This data is then used to train a convolutional neural network that will predict the depth map from the blurry image.

Unsupervised Clustering Based on Pretext Tasks

The idea of this project is that we have two types of neural networks that work together: There is one network A that assigns images to k clusters and k (simple) networks of type B perform a self-supervised task on those clusters. The goal of all the networks is to make the k networks of type B perform well on the task. The assumption is that clustering in semantically similar groups will help the networks of type B to perform well. This could be done on the MNIST dataset with B being linear classifiers and the task being rotation prediction.

Adversarial Data-Augmentation

The student designs a data augmentation network that transforms training images in such a way that image realism is preserved (e.g. with a constrained spatial transformer network) and the transformed images are more difficult to classify (trained via adversarial loss against an image classifier). The model will be evaluated for different data settings (especially in the low data regime), for example on the MNIST and CIFAR datasets.

Unsupervised Learning of Lip-reading from Videos

People with sensory impairment (hearing, speech, vision) depend heavily on assistive technologies to communicate and navigate in everyday life. The mass production of media content today makes it impossible to manually translate everything into a common language for assistive technologies, e.g. captions or sign language.  In this project, the student employs a neural network to learn a representation for lip-movement in videos in an unsupervised fashion, possibly with an encoder-decoder structure where the decoder reconstructs the audio signal. This requires collecting a large dataset of videos (e.g. from YouTube) of speakers or conversations where lip movement is visible. The outcome will be a neural network that learns an audio-visual representation of lip movement in videos, which can then be leveraged to generate captions for hearing impaired persons.

Learning to Generate Topographic Maps from Satellite Images

Satellite images have many applications, e.g. in meteorology, geography, education, cartography and warfare. They are an accurate and detailed depiction of the surface of the earth from above. Although it is relatively simple to collect many satellite images in an automated way, challenges arise when processing them for use in navigation and cartography. The idea of this project is to automatically convert an arbitrary satellite image, of e.g. a city, to a map of simple 2D shapes (streets, houses, forests) and label them with colors (semantic segmentation). The student will collect a dataset of satellite image and topological maps and train a deep neural network that learns to map from one domain to the other. The data could be obtained from a Google Maps database or similar.

New Variables of Brain Morphometry: the Potential and Limitations of CNN Regression

Timo blattner · sept. 2022.

The calculation of variables of brain morphology is computationally very expensive and time-consuming. A previous work showed the feasibility of ex- tracting the variables directly from T1-weighted brain MRI images using a con- volutional neural network. We used significantly more data and extended their model to a new set of neuromorphological variables, which could become inter- esting biomarkers in the future for the diagnosis of brain diseases. The model shows for nearly all subjects a less than 5% mean relative absolute error. This high relative accuracy can be attributed to the low morphological variance be- tween subjects and the ability of the model to predict the cortical atrophy age trend. The model however fails to capture all the variance in the data and shows large regional differences. We attribute these limitations in part to the moderate to poor reliability of the ground truth generated by FreeSurfer. We further investigated the effects of training data size and model complexity on this regression task and found that the size of the dataset had a significant impact on performance, while deeper models did not perform better. Lack of interpretability and dependence on a silver ground truth are the main drawbacks of this direct regression approach.

Home Monitoring by Radar

Lars ziegler · sept. 2022.

Detection and tracking of humans via UWB radars is a promising and continuously evolving field with great potential for medical technology. This contactless method of acquiring data of a patients movement patterns is ideal for in home application. As irregularities in a patients movement patterns are an indicator for various health problems including neurodegenerative diseases, the insight this data could provide may enable earlier detection of such problems. In this thesis a signal processing pipeline is presented with which a persons movement is modeled. During an experiment 142 measurements were recorded by two separate radar systems and one lidar system which each consisted of multiple sensors. The models that were calculated on these measurements by the signal processing pipeline were used to predict the times when a person stood up or sat down. The predictions showed an accuracy of 72.2%.

Revisiting non-learning based 3D reconstruction from multiple images

Aaron sägesser · oct. 2021.

Arthroscopy consists of challenging tasks and requires skills that even today, young surgeons still train directly throughout the surgery. Existing simulators are expensive and rarely available. Through the growing potential of virtual reality(VR) (head-mounted) devices for simulation and their applicability in the medical context, these devices have become a promising alternative that would be orders of magnitude cheaper and could be made widely available. To build a VR-based training device for arthroscopy is the overall aim of our project, as this would be of great benefit and might even be applicable in other minimally invasive surgery (MIS). This thesis marks a first step of the project with its focus to explore and compare well-known algorithms in a multi-view stereo (MVS) based 3D reconstruction with respect to imagery acquired by an arthroscopic camera. Simultaneously with this reconstruction, we aim to gain essential measures to compare the VR environment to the real world, as validation of the realism of future VR tasks. We evaluate 3 different feature extraction algorithms with 3 different matching techniques and 2 different algorithms for the estimation of the fundamental (F) matrix. The evaluation of these 18 different setups is made with a reconstruction pipeline embedded in a jupyter notebook implemented in python based on common computer vision libraries and compared with imagery generated with a mobile phone as well as with the reconstruction results of state-of-the-art (SOTA) structure-from-motion (SfM) software COLMAP and Multi-View Environment (MVE). Our comparative analysis manifests the challenges of heavy distortion, the fish-eye shape and weak image quality of arthroscopic imagery, as all results are substantially worse using this data. However, there are huge differences regarding the different setups. Scale Invariant Feature Transform (SIFT) and Oriented FAST Rotated BRIEF (ORB) in combination with k-Nearest Neighbour (kNN) matching and Least Median of Squares (LMedS) present the most promising results. Overall, the 3D reconstruction pipeline is a useful tool to foster the process of gaining measurements from the arthroscopic exploration device and to complement the comparative research in this context.

Examination of Unsupervised Representation Learning by Predicting Image Rotations

Eric lagger · sept. 2020.

In recent years deep convolutional neural networks achieved a lot of progress. To train such a network a lot of data is required and in supervised learning algorithms it is necessary that the data is labeled. To label data there is a lot of human work needed and this takes a lot of time and money to be done. To avoid the inconveniences that come with this we would like to find systems that don’t need labeled data and therefore are unsupervised learning algorithms. This is the importance of unsupervised algorithms, even though their outcome is not yet on the same qualitative level as supervised algorithms. In this thesis we will discuss an approach of such a system and compare the results to other papers. A deep convolutional neural network is trained to learn the rotations that have been applied to a picture. So we take a large amount of images and apply some simple rotations and the task of the network is to discover in which direction the image has been rotated. The data doesn’t need to be labeled to any category or anything else. As long as all the pictures are upside down we hope to find some high dimensional patterns for the network to learn.

StitchNet: Image Stitching using Autoencoders and Deep Convolutional Neural Networks

Maurice rupp · sept. 2019.

This thesis explores the prospect of artificial neural networks for image processing tasks. More specifically, it aims to achieve the goal of stitching multiple overlapping images to form a bigger, panoramic picture. Until now, this task is solely approached with ”classical”, hardcoded algorithms while deep learning is at most used for specific subtasks. This thesis introduces a novel end-to-end neural network approach to image stitching called StitchNet, which uses a pre-trained autoencoder and deep convolutional networks. Additionally to presenting several new datasets for the task of supervised image stitching with each 120’000 training and 5’000 validation samples, this thesis also conducts various experiments with different kinds of existing networks designed for image superresolution and image segmentation adapted to the task of image stitching. StitchNet outperforms most of the adapted networks in both quantitative as well as qualitative results.

Facial Expression Recognition in the Wild

Luca rolshoven · sept. 2019.

The idea of inferring the emotional state of a subject by looking at their face is nothing new. Neither is the idea of automating this process using computers. Researchers used to computationally extract handcrafted features from face images that had proven themselves to be effective and then used machine learning techniques to classify the facial expressions using these features. Recently, there has been a trend towards using deeplearning and especially Convolutional Neural Networks (CNNs) for the classification of these facial expressions. Researchers were able to achieve good results on images that were taken in laboratories under the same or at least similar conditions. However, these models do not perform very well on more arbitrary face images with different head poses and illumination. This thesis aims to show the challenges of Facial Expression Recognition (FER) in this wild setting. It presents the currently used datasets and the present state-of-the-art results on one of the biggest facial expression datasets currently available. The contributions of this thesis are twofold. Firstly, I analyze three famous neural network architectures and their effectiveness on the classification of facial expressions. Secondly, I present two modifications of one of these networks that lead to the proposed STN-COV model. While this model does not outperform all of the current state-of-the-art models, it does beat several ones of them.

A Study of 3D Reconstruction of Varying Objects with Deformable Parts Models

Raoul grossenbacher · july 2019.

This work covers a new approach to 3D reconstruction. In traditional 3D reconstruction one uses multiple images of the same object to calculate a 3D model by taking information gained from the differences between the images, like camera position, illumination of the images, rotation of the object and so on, to compute a point cloud representing the object. The characteristic trait shared by all these approaches is that one can almost change everything about the image, but it is not possible to change the object itself, because one needs to find correspondences between the images. To be able to use different instances of the same object, we used a 3D DPM model that can find different parts of an object in an image, thereby detecting the correspondences between the different pictures, which we then can use to calculate the 3D model. To take this theory to practise, we gave a 3D DPM model, which was trained to detect cars, pictures of different car brands, where no pair of images showed the same vehicle and used the detected correspondences and the Factorization Method to compute the 3D point cloud. This technique leads to a completely new approach in 3D reconstruction, because changing the object itself was never done before.

Motion deblurring in the wild replication and improvements

Alvaro juan lahiguera · jan. 2019, coma outcome prediction with convolutional neural networks, stefan jonas · oct. 2018, automatic correction of self-introduced errors in source code, sven kellenberger · aug. 2018, neural face transfer: training a deep neural network to face-swap, till nikolaus schnabel · july 2018.

This thesis explores the field of artificial neural networks with realistic looking visual outputs. It aims at morphing face pictures of a specific identity to look like another individual by only modifying key features, such as eye color, while leaving identity-independent features unchanged. Prior works have covered the topic of symmetric translation between two specific domains but failed to optimize it on faces where only parts of the image may be changed. This work applies a face masking operation to the output at training time, which forces the image generator to preserve colors while altering the face, fitting it naturally inside the unmorphed surroundings. Various experiments are conducted including an ablation study on the final setting, decreasing the baseline identity switching performance from 81.7% to 75.8 % whilst improving the average χ2 color distance from 0.551 to 0.434. The provided code-based software gives users easy access to apply this neural face swap to images and videos of arbitrary crop and brings Computer Vision one step closer to replacing Computer Graphics in this specific area.

A Study of the Importance of Parts in the Deformable Parts Model

Sammer puran · june 2017, self-similarity as a meta feature, lucas husi · april 2017, a study of 3d deformable parts models for detection and pose-estimation, simon jenni · march 2015, accelerated federated learning on client silos with label noise: rho selection in classification and segmentation, irakli kelbakiani · may 2024.

Federated Learning has recently gained more research interest. This increased attention is caused by factors including the growth of decentralized data, privacy concerns, and new privacy regulations. In Federated Learning, remote servers keep training a model on local datasets independently, and subsequently, local models are aggregated into a global model, which achieves better overall performance. Sending local model weights instead of the entire dataset is a significant advantage of Federated Learning over centralized classical machine learning algorithms. Federated learning involves uploading and downloading model parameters multiple times, so there are multiple communication rounds between the global server and remote client servers, which imposes challenges. The high number of necessary communication rounds not only increases high-cost communication overheads but is also a critical limitation for servers with low network bandwidth, which leads to latency and a higher probability of training failures caused by communication breakdowns. To mitigate these challenges, we aim to provide a fast-convergent Federated Learning training methodology that decreases the number of necessary communication rounds. We found a paper about Reducible Holdout Loss Selection (RHO-Loss) batch selection methodology, which ”selects low-noise, task-relevant, non-redundant points for training” [1]; we hypothesize, if client silos employ RHO-Loss methodology and successfully avoid training their local models on noisy and non-relevant samples, clients may offer stable and consistent updates to the global server, which could lead to faster convergence of the global model. Our contribution focuses on investigating the RHO-Loss method in a simulated federated setting for the Clothing1M dataset. We also examine its applicability to medical datasets and check its effectiveness in a simulated federated environment. Our experimental results show a promising outcome, specifically a reduction in communication rounds for the Clothing1M dataset. However, as the success of the RHO-Loss selection method depends on the availability of sufficient training data for the target RHO model and for the Irreducible RHO model, we emphasize that our contribution applies to those Federated Learning scenarios where client silos hold enough training data to successfully train and benefit from their RHO model on their local dataset.

Amodal Leaf Segmentation

Nicolas maier · nov. 2023.

Plant phenotyping is the process of measuring and analyzing various traits of plants. It provides essential information on how genetic and environmental factors affect plant growth and development. Manual phenotyping is highly time-consuming; therefore, many computer vision and machine learning based methods have been proposed in the past years to perform this task automatically based on images of the plants. However, the publicly available datasets (in particular, of Arabidopsis thaliana) are limited in size and diversity, making them unsuitable to generalize to new unseen environments. In this work, we propose a complete pipeline able to automatically extract traits of interest from an image of Arabidopsis thaliana. Our method uses a minimal amount of existing annotated data from a source domain to generate a large synthetic dataset adapted to a different target domain (e.g., different backgrounds, lighting conditions, and plant layouts). In addition, unlike the source dataset, the synthetic one provides ground-truth annotations for the occluded parts of the leaves, which are relevant when measuring some characteristics of the plant, e.g., its total area. This synthetic dataset is then used to train a model to perform amodal instance segmentation of the leaves to obtain the total area, leaf count, and color of each plant. To validate our approach, we create a small dataset composed of manually annotated real images of Arabidopsis thaliana, which is used to assess the performance of the models.

Assessment of movement and pose in a hospital bed by ambient and wearable sensor technology in healthy subjects

Tony licata · sept. 2022.

The use of automated systems describing the human motion has become possible in various domains. Most of the proposed systems are designed to work with people moving around in a standing position. Because such system could be interesting in a medical environment, we propose in this work a pipeline that can effectively predict human motion from people lying on beds. The proposed pipeline is tested with a data set composed of 41 participants executing 7 predefined tasks in a bed. The motion of the participants is measured with video cameras, accelerometers and pressure mat. Various experiments are carried with the information retrieved from the data set. Two approaches combining the data from the different measure technologies are explored. The performance of the different carried experiments is measured, and the proposed pipeline is composed with components providing the best results. Later on, we show that the proposed pipeline only needs to use the video cameras, which make the proposed environment easier to implement in real life situations.

Machine Learning Based Prediction of Mental Health Using Wearable-measured Time Series

Seyedeh sharareh mirzargar · sept. 2022.

Depression is the second major cause for years spent in disability and has a growing prevalence in adolescents. The recent Covid-19 pandemic has intensified the situation and limited in-person patient monitoring due to distancing measures. Recent advances in wearable devices have made it possible to record the rest/activity cycle remotely with high precision and in real-world contexts. We aim to use machine learning methods to predict an individual's mental health based on wearable-measured sleep and physical activity. Predicting an impending mental health crisis of an adolescent allows for prompt intervention, detection of depression onset or its recursion, and remote monitoring. To achieve this goal, we train three primary forecasting models; linear regression, random forest, and light gradient boosted machine (LightGBM); and two deep learning models; block recurrent neural network (block RNN) and temporal convolutional network (TCN); on Actigraph measurements to forecast mental health in terms of depression, anxiety, sleepiness, stress, sleep quality, and behavioral problems. Our models achieve a high forecasting performance, the random forest being the winner to reach an accuracy of 98% for forecasting the trait anxiety. We perform extensive experiments to evaluate the models' performance in accuracy, generalization, and feature utilization, using a naive forecaster as the baseline. Our analysis shows minimal mental health changes over two months, making the prediction task easily achievable. Due to these minimal changes in mental health, the models tend to primarily use the historical values of mental health evaluation instead of Actigraph features. At the time of this master thesis, the data acquisition step is still in progress. In future work, we plan to train the models on the complete dataset using a longer forecasting horizon to increase the level of mental health changes and perform transfer learning to compensate for the small dataset size. This interdisciplinary project demonstrates the opportunities and challenges in machine learning based prediction of mental health, paving the way toward using the same techniques to forecast other mental disorders such as internalizing disorder, Parkinson's disease, Alzheimer's disease, etc. and improving the quality of life for individuals who have some mental disorder.

CNN Spike Detector: Detection of Spikes in Intracranial EEG using Convolutional Neural Networks

Stefan jonas · oct. 2021.

The detection of interictal epileptiform discharges in the visual analysis of electroencephalography (EEG) is an important but very difficult, tedious, and time-consuming task. There have been decades of research on computer-assisted detection algorithms, most recently focused on using Convolutional Neural Networks (CNNs). In this thesis, we present the CNN Spike Detector, a convolutional neural network to detect spikes in intracranial EEG. Our dataset of 70 intracranial EEG recordings from 26 subjects with epilepsy introduces new challenges in this research field. We report cross-validation results with a mean AUC of 0.926 (+- 0.04), an area under the precision-recall curve (AUPRC) of 0.652 (+- 0.10) and 12.3 (+- 7.47) false positive epochs per minute for a sensitivity of 80%. A visual examination of false positive segments is performed to understand the model behavior leading to a relatively high false detection rate. We notice issues with the evaluation measures and highlight a major limitation of the common approach of detecting spikes using short segments, namely that the network is not capable to consider the greater context of the segment with regards to its origination. For this reason, we present the Context Model, an extension in which the CNN Spike Detector is supplied with additional information about the channel. Results show promising but limited performance improvements. This thesis provides important findings about the spike detection task for intracranial EEG and lays out promising future research directions to develop a network capable of assisting experts in real-world clinical applications.

PolitBERT - Deepfake Detection of American Politicians using Natural Language Processing

Maurice rupp · april 2021.

This thesis explores the application of modern Natural Language Processing techniques to the detection of artificially generated videos of popular American politicians. Instead of focusing on detecting anomalies and artifacts in images and sounds, this thesis focuses on detecting irregularities and inconsistencies in the words themselves, opening up a new possibility to detect fake content. A novel, domain-adapted, pre-trained version of the language model BERT combined with several mechanisms to overcome severe dataset imbalances yielded the best quantitative as well as qualitative results. Additionally to the creation of the biggest publicly available dataset of English-speaking politicians consisting of 1.5 M sentences from over 1000 persons, this thesis conducts various experiments with different kinds of text classification and sequence processing algorithms applied to the political domain. Furthermore, multiple ablations to manage severe data imbalance are presented and evaluated.

A Study on the Inversion of Generative Adversarial Networks

Ramona beck · march 2021.

The desire to use generative adversarial networks (GANs) for real-world tasks such as object segmentation or image manipulation is increasing as synthesis quality improves, which has given rise to an emerging research area called GAN inversion that focuses on exploring methods for embedding real images into the latent space of a GAN. In this work, we investigate different GAN inversion approaches using an existing generative model architecture that takes a completely unsupervised approach to object segmentation and is based on StyleGAN2. In particular, we propose and analyze algorithms for embedding real images into the different latent spaces Z, W, and W+ of StyleGAN following an optimization-based inversion approach, while also investigating a novel approach that allows fine-tuning of the generator during the inversion process. Furthermore, we investigate a hybrid and a learning-based inversion approach, where in the former we train an encoder with embeddings optimized by our best optimization-based inversion approach, and in the latter we define an autoencoder, consisting of an encoder and the generator of our generative model as a decoder, and train it to map an image into the latent space. We demonstrate the effectiveness of our methods as well as their limitations through a quantitative comparison with existing inversion methods and by conducting extensive qualitative and quantitative experiments with synthetic data as well as real images from a complex image dataset. We show that we achieve qualitatively satisfying embeddings in the W and W+ spaces with our optimization-based algorithms, that fine-tuning the generator during the inversion process leads to qualitatively better embeddings in all latent spaces studied, and that the learning-based approach also benefits from a variable generator as well as a pre-training with our hybrid approach. Furthermore, we evaluate our approaches on the object segmentation task and show that both our optimization-based and our hybrid and learning-based methods are able to generate meaningful embeddings that achieve reasonable object segmentations. Overall, our proposed methods illustrate the potential that lies in the GAN inversion and its application to real-world tasks, especially in the relaxed version of the GAN inversion where the weights of the generator are allowed to vary.

Multi-scale Momentum Contrast for Self-supervised Image Classification

Zhao xueqi · dec. 2020.

With the maturity of supervised learning technology, people gradually shift the research focus to the field of self-supervised learning. ”Momentum Contrast” (MoCo) proposes a new self-supervised learning method and raises the correct rate of self-supervised learning to a new level. Inspired by another article ”Representation Learning by Learning to Count”, if a picture is divided into four parts and passed through a neural network, it is possible to further improve the accuracy of MoCo. Different from the original MoCo, this MoCo variant (Multi-scale MoCo) does not directly pass the image through the encoder after the augmented images. Multi-scale MoCo crops and resizes the augmented images, and the obtained four parts are respectively passed through the encoder and then summed (upsampled version do not do resize to input but resize the contrastive samples). This method of images crop is not only used for queue q but also used for comparison queue k, otherwise the weights of queue k might be damaged during the moment update. This will further discussed in the experiments chapter between downsampled Multi-scale version and downsampled both Multi-scale version. Human beings also have the same principle of object recognition: when human beings see something they are familiar with, even if the object is not fully displayed, people can still guess the object itself with a high probability. Because of this, Multi-scale MoCo applies this concept to the pretext part of MoCo, hoping to obtain better feature extraction. In this thesis, there are three versions of Multi-scale MoCo, downsampled input samples version, downsampled input samples and contrast samples version and upsampled input samples version. The differences between these versions will be described in more detail later. The neural network architecture comparison includes ResNet50 , and the tested data set is STL-10. The weights obtained in pretext will be transferred to self-supervised learning, and in the process of self-supervised learning, the weights of other layers except the final linear layer are frozen without changing (these weights come from pretext).

Self-Supervised Learning Using Siamese Networks and Binary Classifier

Dušan mihajlov · march 2020.

In this thesis, we present several approaches for training a convolutional neural network using only unlabeled data. Our autonomously supervised learning algorithms are based on connections between image patch i. e. zoomed image and its original. Using the siamese architecture neural network we aim to recognize, if the image patch, which is input to the first neural network part, comes from the same image presented to the second neural network part. By applying transformations to both images, and different zoom sizes at different positions, we force the network to extract high level features using its convolutional layers. At the top of our siamese architecture, we have a simple binary classifier that measures the difference between feature maps that we extract and makes a decision. Thus, the only way that the classifier will solve the task correctly is when our convolutional layers are extracting useful representations. Those representations we can than use to solve many different tasks that are related to the data used for unsupervised training. As the main benchmark for all of our models, we used STL10 dataset, where we train a linear classifier on the top of our convolutional layers with a small amount of manually labeled images, which is a widely used benchmark for unsupervised learning tasks. We also combine our idea with recent work on the same topic, and the network called RotNet, which makes use of image rotations and therefore forces the network to learn rotation dependent features from the dataset. As a result of this combination we create a new procedure that outperforms original RotNet.

Learning Object Representations by Mixing Scenes

Lukas zbinden · may 2019.

In the digital age of ever increasing data amassment and accessibility, the demand for scalable machine learning models effective at refining the new oil is unprecedented. Unsupervised representation learning methods present a promising approach to exploit this invaluable yet unlabeled digital resource at scale. However, a majority of these approaches focuses on synthetic or simplified datasets of images. What if a method could learn directly from natural Internet-scale image data? In this thesis, we propose a novel approach for unsupervised learning of object representations by mixing natural image scenes. Without any human help, our method mixes visually similar images to synthesize new realistic scenes using adversarial training. In this process the model learns to represent and understand the objects prevalent in natural image data and makes them available for downstream applications. For example, it enables the transfer of objects from one scene to another. Through qualitative experiments on complex image data we show the effectiveness of our method along with its limitations. Moreover, we benchmark our approach quantitatively against state-of-the-art works on the STL-10 dataset. Our proposed method demonstrates the potential that lies in learning representations directly from natural image data and reinforces it as a promising avenue for future research.

Representation Learning using Semantic Distances

Markus roth · may 2019, zero-shot learning using generative adversarial networks, hamed hemati · dec. 2018, dimensionality reduction via cnns - learning the distance between images, ioannis glampedakis · sept. 2018, learning to play othello using deep reinforcement learning and self play, thomas simon steinmann · sept. 2018, aba-j interactive multi-modality tissue sectionto-volume alignment: a brain atlasing toolkit for imagej, felix meyenhofer · march 2018, learning visual odometry with recurrent neural networks, adrian wälchli · feb. 2018.

In computer vision, Visual Odometry is the problem of recovering the camera motion from a video. It is related to Structure from Motion, the problem of reconstructing the 3D geometry from a collection of images. Decades of research in these areas have brought successful algorithms that are used in applications like autonomous navigation, motion capture, augmented reality and others. Despite the success of these prior works in real-world environments, their robustness is highly dependent on manual calibration and the magnitude of noise present in the images in form of, e.g., non-Lambertian surfaces, dynamic motion and other forms of ambiguity. This thesis explores an alternative approach to the Visual Odometry problem via Deep Learning, that is, a specific form of machine learning with artificial neural networks. It describes and focuses on the implementation of a recent work that proposes the use of Recurrent Neural Networks to learn dependencies over time due to the sequential nature of the input. Together with a convolutional neural network that extracts motion features from the input stream, the recurrent part accumulates knowledge from the past to make camera pose estimations at each point in time. An analysis on the performance of this system is carried out on real and synthetic data. The evaluation covers several ways of training the network as well as the impact and limitations of the recurrent connection for Visual Odometry.

Crime location and timing prediction

Bernard swart · jan. 2018, from cartoons to real images: an approach to unsupervised visual representation learning, simon jenni · feb. 2017, automatic and large-scale assessment of fluid in retinal oct volume, nina mujkanovic · dec. 2016, segmentation in 3d using eye-tracking technology, michele wyss · july 2016, accurate scale thresholding via logarithmic total variation prior, remo diethelm · aug. 2014, novel techniques for robust and generalizable machine learning, abdelhak lemkhenter · sept. 2023.

Neural networks have transcended their status of powerful proof-of-concept machine learning into the realm of a highly disruptive technology that has revolutionized many quantitative fields such as drug discovery, autonomous vehicles, and machine translation. Today, it is nearly impossible to go a single day without interacting with a neural network-powered application. From search engines to on-device photo-processing, neural networks have become the go-to solution thanks to recent advances in computational hardware and an unprecedented scale of training data. Larger and less curated datasets, typically obtained through web crawling, have greatly propelled the capabilities of neural networks forward. However, this increase in scale amplifies certain challenges associated with training such models. Beyond toy or carefully curated datasets, data in the wild is plagued with biases, imbalances, and various noisy components. Given the larger size of modern neural networks, such models run the risk of learning spurious correlations that fail to generalize beyond their training data. This thesis addresses the problem of training more robust and generalizable machine learning models across a wide range of learning paradigms for medical time series and computer vision tasks. The former is a typical example of a low signal-to-noise ratio data modality with a high degree of variability between subjects and datasets. There, we tailor the training scheme to focus on robust patterns that generalize to new subjects and ignore the noisier and subject-specific patterns. To achieve this, we first introduce a physiologically inspired unsupervised training task and then extend it by explicitly optimizing for cross-dataset generalization using meta-learning. In the context of image classification, we address the challenge of training semi-supervised models under class imbalance by designing a novel label refinement strategy with higher local sensitivity to minority class samples while preserving the global data distribution. Lastly, we introduce a new Generative Adversarial Networks training loss. Such generative models could be applied to improve the training of subsequent models in the low data regime by augmenting the dataset using generated samples. Unfortunately, GAN training relies on a delicate balance between its components, making it prone mode collapse. Our contribution consists of defining a more principled GAN loss whose gradients incentivize the generator model to seek out missing modes in its distribution. All in all, this thesis tackles the challenge of training more robust machine learning models that can generalize beyond their training data. This necessitates the development of methods specifically tailored to handle the diverse biases and spurious correlations inherent in the data. It is important to note that achieving greater generalizability in models goes beyond simply increasing the volume of data; it requires meticulous consideration of training objectives and model architecture. By tackling these challenges, this research contributes to advancing the field of machine learning and underscores the significance of thoughtful design in obtaining more resilient and versatile models.

Automated Sleep Scoring, Deep Learning and Physician Supervision

Luigi fiorillo · oct. 2022.

Sleep plays a crucial role in human well-being. Polysomnography is used in sleep medicine as a diagnostic tool, so as to objectively analyze the quality of sleep. Sleep scoring is the procedure of extracting sleep cycle information from the wholenight electrophysiological signals. The scoring is done worldwide by the sleep physicians according to the official American Academy of Sleep Medicine (AASM) scoring manual. In the last decades, a wide variety of deep learning based algorithms have been proposed to automatise the sleep scoring task. In this thesis we study the reasons why these algorithms fail to be introduced in the daily clinical routine, with the perspective of bridging the existing gap between the automatic sleep scoring models and the sleep physicians. In this light, the primary step is the design of a simplified sleep scoring architecture, also providing an estimate of the model uncertainty. Beside achieving results on par with most up-to-date scoring systems, we demonstrate the efficiency of ensemble learning based algorithms, together with label smoothing techniques, in both enhancing the performance and calibrating the simplified scoring model. We introduced an uncertainty estimate procedure, so as to identify the most challenging sleep stage predictions, and to quantify the disagreement between the predictions given by the model and the annotation given by the physicians. In this thesis we also propose a novel method to integrate the inter-scorer variability into the training procedure of a sleep scoring model. We clearly show that a deep learning model is able to encode this variability, so as to better adapt to the consensus of a group of scorers-physicians. We finally address the generalization ability of a deep learning based sleep scoring system, further studying its resilience to the sleep complexity and to the AASM scoring rules. We can state that there is no need to train the algorithm strictly following the AASM guidelines. Most importantly, using data from multiple data centers results in a better performing model compared with training on a single data cohort. The variability among different scorers and data centers needs to be taken into account, more than the variability among sleep disorders.

Learning Representations for Controllable Image Restoration

Givi meishvili · march 2022.

Deep Convolutional Neural Networks have sparked a renaissance in all the sub-fields of computer vision. Tremendous progress has been made in the area of image restoration. The research community has pushed the boundaries of image deblurring, super-resolution, and denoising. However, given a distorted image, most existing methods typically produce a single restored output. The tasks mentioned above are inherently ill-posed, leading to an infinite number of plausible solutions. This thesis focuses on designing image restoration techniques capable of producing multiple restored results and granting users more control over the restoration process. Towards this goal, we demonstrate how one could leverage the power of unsupervised representation learning. Image restoration is vital when applied to distorted images of human faces due to their social significance. Generative Adversarial Networks enable an unprecedented level of generated facial details combined with smooth latent space. We leverage the power of GANs towards the goal of learning controllable neural face representations. We demonstrate how to learn an inverse mapping from image space to these latent representations, tuning these representations towards a specific task, and finally manipulating latent codes in these spaces. For example, we show how GANs and their inverse mappings enable the restoration and editing of faces in the context of extreme face super-resolution and the generation of novel view sharp videos from a single motion-blurred image of a face. This thesis also addresses more general blind super-resolution, denoising, and scratch removal problems, where blur kernels and noise levels are unknown. We resort to contrastive representation learning and first learn the latent space of degradations. We demonstrate that the learned representation allows inference of ground-truth degradation parameters and can guide the restoration process. Moreover, it enables control over the amount of deblurring and denoising in the restoration via manipulation of latent degradation features.

Learning Generalizable Visual Patterns Without Human Supervision

Simon jenni · oct. 2021.

Owing to the existence of large labeled datasets, Deep Convolutional Neural Networks have ushered in a renaissance in computer vision. However, almost all of the visual data we generate daily - several human lives worth of it - remains unlabeled and thus out of reach of today’s dominant supervised learning paradigm. This thesis focuses on techniques that steer deep models towards learning generalizable visual patterns without human supervision. Our primary tool in this endeavor is the design of Self-Supervised Learning tasks, i.e., pretext-tasks for which labels do not involve human labor. Besides enabling the learning from large amounts of unlabeled data, we demonstrate how self-supervision can capture relevant patterns that supervised learning largely misses. For example, we design learning tasks that learn deep representations capturing shape from images, motion from video, and 3D pose features from multi-view data. Notably, these tasks’ design follows a common principle: The recognition of data transformations. The strong performance of the learned representations on downstream vision tasks such as classification, segmentation, action recognition, or pose estimation validate this pretext-task design. This thesis also explores the use of Generative Adversarial Networks (GANs) for unsupervised representation learning. Besides leveraging generative adversarial learning to define image transformation for self-supervised learning tasks, we also address training instabilities of GANs through the use of noise. While unsupervised techniques can significantly reduce the burden of supervision, in the end, we still rely on some annotated examples to fine-tune learned representations towards a target task. To improve the learning from scarce or noisy labels, we describe a supervised learning algorithm with improved generalization in these challenging settings.

Learning Interpretable Representations of Images

Attila szabó · june 2019.

Computers represent images with pixels and each pixel contains three numbers for red, green and blue colour values. These numbers are meaningless for humans and they are mostly useless when used directly with classical machine learning techniques like linear classifiers. Interpretable representations are the attributes that humans understand: the colour of the hair, viewpoint of a car or the 3D shape of the object in the scene. Many computer vision tasks can be viewed as learning interpretable representations, for example a supervised classification algorithm directly learns to represent images with their class labels. In this work we aim to learn interpretable representations (or features) indirectly with lower levels of supervision. This approach has the advantage of cost savings on dataset annotations and the flexibility of using the features for multiple follow-up tasks. We made contributions in three main areas: weakly supervised learning, unsupervised learning and 3D reconstruction. In the weakly supervised case we use image pairs as supervision. Each pair shares a common attribute and differs in a varying attribute. We propose a training method that learns to separate the attributes into separate feature vectors. These features then are used for attribute transfer and classification. We also show theoretical results on the ambiguities of the learning task and the ways to avoid degenerate solutions. We show a method for unsupervised representation learning, that separates semantically meaningful concepts. We explain and show ablation studies how the components of our proposed method work: a mixing autoencoder, a generative adversarial net and a classifier. We propose a method for learning single image 3D reconstruction. It is done using only the images, no human annotation, stereo, synthetic renderings or ground truth depth map is needed. We train a generative model that learns the 3D shape distribution and an encoder to reconstruct the 3D shape. For that we exploit the notion of image realism. It means that the 3D reconstruction of the object has to look realistic when it is rendered from different random angles. We prove the efficacy of our method from first principles.

Learning Controllable Representations for Image Synthesis

Qiyang hu · june 2019.

In this thesis, our focus is learning a controllable representation and applying the learned controllable feature representation on images synthesis, video generation, and even 3D reconstruction. We propose different methods to disentangle the feature representation in neural network and analyze the challenges in disentanglement such as reference ambiguity and shortcut problem when using the weak label. We use the disentangled feature representation to transfer attributes between images such as exchanging hairstyle between two face images. Furthermore, we study the problem of how another type of feature, sketch, works in a neural network. The sketch can provide shape and contour of an object such as the silhouette of the side-view face. We leverage the silhouette constraint to improve the 3D face reconstruction from 2D images. The sketch can also provide the moving directions of one object, thus we investigate how one can manipulate the object to follow the trajectory provided by a user sketch. We propose a method to automatically generate video clips from a single image input using the sketch as motion and trajectory guidance to animate the object in that image. We demonstrate the efficiency of our approaches on several synthetic and real datasets.

Beyond Supervised Representation Learning

Mehdi noroozi · jan. 2019.

The complexity of any information processing task is highly dependent on the space where data is represented. Unfortunately, pixel space is not appropriate for the computer vision tasks such as object classification. The traditional computer vision approaches involve a multi-stage pipeline where at first images are transformed to a feature space through a handcrafted function and then consequenced by the solution in the feature space. The challenge with this approach is the complexity of designing handcrafted functions that extract robust features. The deep learning based approaches address this issue by end-to-end training of a neural network for some tasks that lets the network to discover the appropriate representation for the training tasks automatically. It turns out that image classification task on large scale annotated datasets yields a representation transferable to other computer vision tasks. However, supervised representation learning is limited to annotations. In this thesis we study self-supervised representation learning where the goal is to alleviate these limitations by substituting the classification task with pseudo tasks where the labels come for free. We discuss self-supervised learning by solving jigsaw puzzles that uses context as supervisory signal. The rational behind this task is that the network requires to extract features about object parts and their spatial configurations to solve the jigsaw puzzles. We also discuss a method for representation learning that uses an artificial supervisory signal based on counting visual primitives. This supervisory signal is obtained from an equivariance relation. We use two image transformations in the context of counting: scaling and tiling. The first transformation exploits the fact that the number of visual primitives should be invariant to scale. The second transformation allows us to equate the total number of visual primitives in each tile to that in the whole image. The most effective transfer strategy is fine-tuning, which restricts one to use the same model or parts thereof for both pretext and target tasks. We discuss a novel framework for self-supervised learning that overcomes limitations in designing and comparing different tasks, models, and data domains. In particular, our framework decouples the structure of the self-supervised model from the final task-specific finetuned model. Finally, we study the problem of multi-task representation learning. A naive approach to enhance the representation learned by a task is to train the task jointly with other tasks that capture orthogonal attributes. Having a diverse set of auxiliary tasks, imposes challenges on multi-task training from scratch. We propose a framework that allows us to combine arbitrarily different feature spaces into a single deep neural network. We reduce the auxiliary tasks to classification tasks and the multi-task learning to multi-label classification task consequently. Nevertheless, combining multiple representation space without being aware of the target task might be suboptimal. As our second contribution, we show empirically that this is indeed the case and propose to combine multiple tasks after the fine-tuning on the target task.

Motion Deblurring from a Single Image

Meiguang jin · dec. 2018.

With the information explosion, a tremendous amount photos is captured and shared via social media everyday. Technically, a photo requires a finite exposure to accumulate light from the scene. Thus, objects moving during the exposure generate motion blur in a photo. Motion blur is an image degradation that makes visual content less interpretable and is therefore often seen as a nuisance. Although motion blur can be reduced by setting a short exposure time, an insufficient amount of light has to be compensated through increasing the sensor’s sensitivity, which will inevitably bring large amount of sensor noise. Thus this motivates the necessity of removing motion blur computationally. Motion deblurring is an important problem in computer vision and it is challenging due to its ill-posed nature, which means the solution is not well defined. Mathematically, a blurry image caused by uniform motion is formed by the convolution operation between a blur kernel and a latent sharp image. Potentially there are infinite pairs of blur kernel and latent sharp image that can result in the same blurry image. Hence, some prior knowledge or regularization is required to address this problem. Even if the blur kernel is known, restoring the latent sharp image is still difficult as the high frequency information has been removed. Although we can model the uniform motion deblurring problem mathematically, it can only address the camera in-plane translational motion. Practically, motion is more complicated and can be non-uniform. Non-uniform motion blur can come from many sources, camera out-of-plane rotation, scene depth change, object motion and so on. Thus, it is more challenging to remove non-uniform motion blur. In this thesis, our focus is motion blur removal. We aim to address four challenging motion deblurring problems. We start from the noise blind image deblurring scenario where blur kernel is known but the noise level is unknown. We introduce an efficient and robust solution based on a Bayesian framework using a smooth generalization of the 0−1 loss to address this problem. Then we study the blind uniform motion deblurring scenario where both the blur kernel and the latent sharp image are unknown. We explore the relative scale ambiguity between the latent sharp image and blur kernel to address this issue. Moreover, we study the face deblurring problem and introduce a novel deep learning network architecture to solve it. We also address the general motion deblurring problem and particularly we aim at recovering a sequence of 7 frames each depicting some instantaneous motion of the objects in the scene.

Towards a Novel Paradigm in Blind Deconvolution: From Natural to Cartooned Image Statistics

Daniele perrone · july 2015.

In this thesis we study the blind deconvolution problem. Blind deconvolution consists in the estimation of a sharp image and a blur kernel from an observed blurry image. Because the blur model admits several solutions it is necessary to devise an image prior that favors the true blur kernel and sharp image. Recently it has been shown that a class of blind deconvolution formulations and image priors has the no-blur solution as global minimum. Despite this shortcoming, algorithms based on these formulations and priors can successfully solve blind deconvolution. In this thesis we show that a suitable initialization can exploit the non-convexity of the problem and yield the desired solution. Based on these conclusions, we propose a novel “vanilla” algorithm stripped of any enhancement typically used in the literature. Our algorithm, despite its simplicity, is able to compete with the top performers on several datasets. We have also investigated a remarkable behavior of a 1998 algorithm, whose formulation has the no-blur solution as global minimum: even when initialized at the no-blur solution, it converges to the correct solution. We show that this behavior is caused by an apparently insignificant implementation strategy that makes the algorithm no longer minimize the original cost functional. We also demonstrate that this strategy improves the results of our “vanilla” algorithm. Finally, we present a study of image priors for blind deconvolution. We provide experimental evidence supporting the recent belief that a good image prior is one that leads to a good blur estimate rather than being a good natural image statistical model. By focusing the attention on the blur estimation alone, we show that good blur estimates can be obtained even when using images quite different from the true sharp image. This allows using image priors, such as those leading to “cartooned” images, that avoid the no-blur solution. By using an image prior that produces “cartooned” images we achieve state-of-the-art results on different publicly available datasets. We therefore suggests a shift of paradigm in blind deconvolution: from modeling natural image statistics to modeling cartooned image statistics.

New Perspectives on Uncalibrated Photometric Stereo

Thoma papadhimitri · june 2014.

This thesis investigates the problem of 3D reconstruction of a scene from 2D images. In particular, we focus on photometric stereo which is a technique that computes the 3D geometry from at least three images taken from the same viewpoint and under different illumination conditions. When the illumination is unknown (uncalibrated photometric stereo) the problem is ambiguous: different combinations of geometry and illumination can generate the same images. First, we solve the ambiguity by exploiting the Lambertian reflectance maxima. These are points defined on curved surfaces where the normals are parallel to the light direction. Then, we propose a solution that can be computed in closed-form and thus very efficiently. Our algorithm is also very robust and yields always the same estimate regardless of the initial ambiguity. We validate our method on real world experiments and achieve state-of-art results. In this thesis we also solve for the first time the uncalibrated photometric stereo problem under the perspective projection model. We show that unlike in the orthographic case, one can uniquely reconstruct the normals of the object and the lights given only the input images and the camera calibration (focal length and image center). We also propose a very efficient algorithm which we validate on synthetic and real world experiments and show that the proposed technique is a generalization of the orthographic case. Finally, we investigate the uncalibrated photometric stereo problem in the case where the lights are distributed near the scene. In this case we propose an alternating minimization technique which converges quickly and overcomes the limitations of prior work that assumes distant illumination. We show experimentally that adopting a near-light model for real world scenes yields very accurate reconstructions.

Available Master's thesis topics in machine learning

Main content.

Here we list topics that are available. You may also be interested in our list of completed Master's theses .

Learning and inference with large Bayesian networks

Most learning and inference tasks with Bayesian networks are NP-hard. Therefore, one often resorts to using different heuristics that do not give any quality guarantees.

Task: Evaluate quality of large-scale learning or inference algorithms empirically.

Advisor: Pekka Parviainen

Sum-product networks

Traditionally, probabilistic graphical models use a graph structure to represent dependencies and independencies between random variables. Sum-product networks are a relatively new type of a graphical model where the graphical structure models computations and not the relationships between variables. The benefit of this representation is that inference (computing conditional probabilities) can be done in linear time with respect to the size of the network.

Potential thesis topics in this area: a) Compare inference speed with sum-product networks and Bayesian networks. Characterize situations when one model is better than the other. b) Learning the sum-product networks is done using heuristic algorithms. What is the effect of approximation in practice?

Bayesian Bayesian networks

The naming of Bayesian networks is somewhat misleading because there is nothing Bayesian in them per se; A Bayesian network is just a representation of a joint probability distribution. One can, of course, use a Bayesian network while doing Bayesian inference. One can also learn Bayesian networks in a Bayesian way. That is, instead of finding an optimal network one computes the posterior distribution over networks.

Task: Develop algorithms for Bayesian learning of Bayesian networks (e.g., MCMC, variational inference, EM)

Large-scale (probabilistic) matrix factorization

The idea behind matrix factorization is to represent a large data matrix as a product of two or more smaller matrices.They are often used in, for example, dimensionality reduction and recommendation systems. Probabilistic matrix factorization methods can be used to quantify uncertainty in recommendations. However, large-scale (probabilistic) matrix factorization is computationally challenging.

Potential thesis topics in this area: a) Develop scalable methods for large-scale matrix factorization (non-probabilistic or probabilistic), b) Develop probabilistic methods for implicit feedback (e.g., recommmendation engine when there are no rankings but only knowledge whether a customer has bought an item)

Bayesian deep learning

Standard deep neural networks do not quantify uncertainty in predictions. On the other hand, Bayesian methods provide a principled way to handle uncertainty. Combining these approaches leads to Bayesian neural networks. The challenge is that Bayesian neural networks can be cumbersome to use and difficult to learn.

The task is to analyze Bayesian neural networks and different inference algorithms in some simple setting.

Deep learning for combinatorial problems

Deep learning is usually applied in regression or classification problems. However, there has been some recent work on using deep learning to develop heuristics for combinatorial optimization problems; see, e.g., [1] and [2].

Task: Choose a combinatorial problem (or several related problems) and develop deep learning methods to solve them.

References: [1] Vinyals, Fortunato and Jaitly: Pointer networks. NIPS 2015. [2] Dai, Khalil, Zhang, Dilkina and Song: Learning Combinatorial Optimization Algorithms over Graphs. NIPS 2017.

Advisors: Pekka Parviainen, Ahmad Hemmati

Estimating the number of modes of an unknown function

Mode seeking considers estimating the number of local maxima of a function f. Sometimes one can find modes by, e.g., looking for points where the derivative of the function is zero. However, often the function is unknown and we have only access to some (possibly noisy) values of the function. 

In topological data analysis,  we can analyze topological structures using persistent homologies. For 1-dimensional signals, this can translate into looking at the birth/death persistence diagram, i.e. the birth and death of connected topological components as we expand the space around each point where we have observed our function. These observations turn out to be closely related to the modes (local maxima) of the function. A recent paper [1] proposed an efficient method for mode seeking.

In this project, the task is to extend the ideas from [1] to get a probabilistic estimate on the number of modes. To this end, one has to use probabilistic methods such as Gaussian processes.

[1] U. Bauer, A. Munk, H. Sieling, and M. Wardetzky. Persistence barcodes versus Kolmogorov signatures: Detecting modes of one-dimensional signals. Foundations of computational mathematics17:1 - 33, 2017.

Advisors:  Pekka Parviainen ,  Nello Blaser

Causal Abstraction Learning

We naturally make sense of the world around us by working out causal relationships between objects and by representing in our minds these objects with different degrees of approximation and detail. Both processes are essential to our understanding of reality, and likely to be fundamental for developing artificial intelligence. The first process may be expressed using the formalism of structural causal models, while the second can be grounded in the theory of causal abstraction.        

This project will consider the problem of learning an abstraction between two given structural causal models. The primary goal will be the development of efficient algorithms able to learn a meaningful abstraction between the given causal models.

Advisor: Fabio Massimo Zennaro

Causal Bandits

"Multi-armed bandit" is an informal name for slot machines, and the formal name of a large class of problems where an agent has to choose an action among a range of possibilities without knowing the ensuing rewards. Multi-armed bandit problems are one of the most essential reinforcement learning problems where an agent is directly faced with an exploitation-exploration trade-off.

This project will consider a class of multi-armed bandits where an agent, upon taking an action, interacts with a causal system. The primary goal will be the development of learning strategies that takes advantage of the underlying causal system in order to learn optimal policies in a shortest amount of time.

Causal Modelling for Battery Manufacturing

Lithium-ion batteries are poised to be one of the most important sources of energy in the near future. Yet, the process of manufacturing these batteries is very hard to model and control. Optimizing the different phases of production to maximize the lifetime of the batteries is a non-trivial challenge since physical models are limited in scope and collecting experimental data is extremely expensive and time-consuming.        

This project will consider the problem of aggregating and analyzing data regarding a few stages in the process of battery manufacturing. The primary goal will be the development of algorithms for transporting and integrating data collected in different contexts, as well as the use of explainable algorithms to interpret them.

Reinforcement Learning for Computer Security

The field of computer security presents a wide variety of challenging problems for artificial intelligence and autonomous agents. Guaranteeing the security of a system against attacks and penetrations by malicious hackers has always been a central concern of this field, and machine learning could now offer a substantial contribution. Security capture-the-flag simulations are particularly well-suited as a testbed for the application and development of reinforcement learning algorithms.

This project will consider the use of reinforcement learning for the preventive purpose of testing systems and discovering vulnerabilities before they can be exploited. The primary goal will be the modelling of capture-the-flag challenges of interest and the development of reinforcement learning algorithms that can solve them.

Approaches to AI Safety

The world and the Internet are more and more populated by artificial autonomous agents carrying out tasks on our behalf. Many of these agents are provided with an objective and they learn their behaviour trying to achieve their objective as best as they can. However, this approach can not guarantee that an agent, while learning its behaviour, will not undertake actions that may have unforeseen and undesirable effects. Research in AI safety tries to design autonomous agent that will behave in a predictable and safe way. 

This project will consider specific problems and novel solution in the domain of AI safety and reinforcement learning. The primary goal will be the development of innovative algorithms and their implementation withing established frameworks.

Reinforcement Learning for Super-modelling

Super-modelling [1] is a technique designed for combining together complex dynamical models: pre-trained models are aggregated with messages and information being exchanged in order synchronize the behavior  of the different modles and produce more accurate and reliable predictions. Super-models are used, for instance, in weather or climate science, where pre-existing models are ensembled together and their states dynamically aggregated to generate more realistic simulations. 

This project will consider how reinforcement learning algorithms may be used to solve the coordination problem among the individual models forming a super-model. The primary goal will be the formulation of the super-modelling problem within the reinforcement learning framework and the study of custom RL algorithms to improve the overall performance of super-models.

[1] Schevenhoven, Francine, et al. "Supermodeling: improving predictions with an ensemble of interacting models." Bulletin of the American Meteorological Society 104.9 (2023): E1670-E1686.

Advisor: Fabio Massimo Zennaro ,  Francine Janneke Schevenhoven

The Topology of Flight Paths

Air traffic data tells us the position, direction, and speed of an aircraft at a given time. In other words, if we restrict our focus to a single aircraft, we are looking at a multivariate time-series. We can visualize the flight path as a curve above earth's surface quite geometrically. Topological data analysis (TDA) provides different methods for analysing the shape of data. Consequently, TDA may help us to extract meaningful features from the air traffic data. Although the typical flight path shapes may not be particularly intriguing, we can attempt to identify more intriguing patterns or “abnormal” manoeuvres, such as aborted landings, go-arounds, or diverts.

Advisor:  Odin Hoff Gardå , Nello Blaser

Automatic hyperparameter selection for isomap

Isomap is a non-linear dimensionality reduction method with two free hyperparameters (number of nearest neighbors and neighborhood radius). Different hyperparameters result in dramatically different embeddings. Previous methods for selecting hyperparameters focused on choosing one optimal hyperparameter. In this project, you will explore the use of persistent homology to find parameter ranges that result in stable embeddings. The project has theoretic and computational aspects.

Advisor: Nello Blaser

Validate persistent homology

Persistent homology is a generalization of hierarchical clustering to find more structure than just the clusters. Traditionally, hierarchical clustering has been evaluated using resampling methods and assessing stability properties. In this project you will generalize these resampling methods to develop novel stability properties that can be used to assess persistent homology. This project has theoretic and computational aspects.

Topological Ancombs quartet

This topic is based on the classical Ancombs quartet and families of point sets with identical 1D persistence ( https://arxiv.org/abs/2202.00577 ). The goal is to generate more interesting datasets using the simulated annealing methods presented in ( http://library.usc.edu.ph/ACM/CHI%202017/1proc/p1290.pdf ). This project is mostly computational.

Persistent homology vectorization with cycle location

There are many methods of vectorizing persistence diagrams, such as persistence landscapes, persistence images, PersLay and statistical summaries. Recently we have designed algorithms to in some cases efficiently detect the location of persistence cycles. In this project, you will vectorize not just the persistence diagram, but additional information such as the location of these cycles. This project is mostly computational with some theoretic aspects.

Divisive covers

Divisive covers are a divisive technique for generating filtered simplicial complexes. They original used a naive way of dividing data into a cover. In this project, you will explore different methods of dividing space, based on principle component analysis, support vector machines and k-means clustering. In addition, you will explore methods of using divisive covers for classification. This project will be mostly computational.

Learning Acquisition Functions for Cost-aware Bayesian Optimization

This is a follow-up project of an earlier Master thesis that developed a novel method for learning Acquisition Functions in Bayesian Optimization through the use of Reinforcement Learning. The goal of this project is to further generalize this method (more general input, learned cost-functions) and apply it to hyperparameter optimization for neural networks.

Advisors: Nello Blaser , Audun Ljone Henriksen

Stable updates

This is a follow-up project of an earlier Master thesis that introduced and studied empirical stability in the context of tree-based models. The goal of this project is to develop stable update methods for deep learning models. You will design sevaral stable methods and empirically compare them (in terms of loss and stability) with a baseline and with one another.

Advisors:  Morten Blørstad , Nello Blaser

Multimodality in Bayesian neural network ensembles

One method to assess uncertainty in neural network predictions is to use dropout or noise generators at prediction time and run every prediction many times. This leads to a distribution of predictions. Informatively summarizing such probability distributions is a non-trivial task and the commonly used means and standard deviations result in the loss of crucial information, especially in the case of multimodal distributions with distinct likely outcomes. In this project, you will analyze such multimodal distributions with mixture models and develop ways to exploit such multimodality to improve training. This project can have theoretical, computational and applied aspects.

Learning a hierarchical metric

Often, labels have defined relationships to each other, for instance in a hierarchical taxonomy. E.g. ImageNet labels are derived from the WordNet graph, and biological species are taxonomically related, and can have similarities depending on life stage, sex, or other properties.

ArcFace is an alternative loss function that aims for an embedding that is more generally useful than softmax. It is commonly used in metric learning/few shot learning cases.

Here, we will develop a metric learning method that learns from data with hierarchical labels. Using multiple ArcFace heads, we will simultaneously learn to place representations to optimize the leaf label as well as intermediate labels on the path from leaf to root of the label tree. Using taxonomically classified plankton image data, we will measure performance as a function of ArcFace parameters (sharpness/temperature and margins -- class-wise or level-wise), and compare the results to existing methods.

Advisor: Ketil Malde ( [email protected] )

Self-supervised object detection in video

One challenge with learning object detection is that in many scenes that stretch off into the distance, annotating small, far-off, or blurred objects is difficult. It is therefore desirable to learn from incompletely annotated scenes, and one-shot object detectors may suffer from incompletely annotated training data.

To address this, we will use a region-propsal algorithm (e.g. SelectiveSearch) to extract potential crops from each frame. Classification will be based on two approaches: a) training based on annotated fish vs random similarly-sized crops without annotations, and b) using a self-supervised method to build a representation for crops, and building a classifier for the extracted regions. The method will be evaluated against one-shot detectors and other training regimes.

If successful, the method will be applied to fish detection and tracking in videos from baited and unbaited underwater traps, and used to estimate abundance of various fish species.

See also: Benettino (2016): https://link.springer.com/chapter/10.1007/978-3-319-48881-3_56

Representation learning for object detection

While traditional classifiers work well with data that is labeled with disjoint classes and reasonably balanced class abundances, reality is often less clean. An alternative is to learn a vectors space embedding that reflects semantic relationships between objects, and deriving classes from this representation. This is especially useful for few-shot classification (ie. very few examples in the training data).

The task here is to extend a modern object detector (e.g. Yolo v8) to output an embedding of the identified object. Instead of a softmax classifier, we can learn the embedding either in a supervised manner (using annotations on frames) by attaching an ArcFace or other supervised metric learning head. Alternatively, the representation can be learned from tracked detections over time using e.g. a contrastive loss function to keep the representation for an object (approximately) constant over time. The performance of the resulting object detector will be measured on underwater videos, targeting species detection and/or indiviual recognition (re-ID).

Time-domain object detection

Object detectors for video are normally trained on still frames, but it is evident (from human experience) that using time domain information is more effective. I.e., it can be hard to identify far-off or occluded objects in still images, but movement in time often reveals them.

Here we will extend a state of the art object detector (e.g. yolo v8) with time domain data. Instead of using a single frame as input, the model will be modified to take a set of frames surrounding the annotated frame as input. Performance will be compared to using single-frame detection.

Large-scale visualization of acoustic data

The Institute of Marine Research has decades of acoustic data collected in various surveys. These data are in the process of being converted to data formats that can be processed and analyzed more easily using packages like Xarray and Dask.

The objective is to make these data more accessible to regular users by providing a visual front end. The user should be able to quickly zoom in and out, perform selection, export subsets, apply various filters and classifiers, and overlay annotations and other relevant auxiliary data.

Learning acoustic target classification from simulation

Broadband echosounders emit a complex signal that spans a large frequency band. Different targets will reflect, absorb, and generate resonance at different amplitudes and frequencies, and it is therefore possible to classify targets at much higher resolution and accuracy than before. Due to the complexity of the received signals, deriving effective profiles that can be used to identify targets is difficult.

Here we will use simulated frequency spectra from geometric objects with various shapes, orientation, and other properties. We will train ML models to estimate (recover) the geometric and material properties of objects based on these spectra. The resulting model will be applied to read broadband data, and compared to traditional classification methods.

Online learning in real-time systems

Build a model for the drilling process by using the Virtual simulator OpenLab ( https://openlab.app/ ) for real-time data generation and online learning techniques. The student will also do a short survey of existing online learning techniques and learn how to cope with errors and delays in the data.

Advisor: Rodica Mihai

Building a finite state automaton for the drilling process by using queries and counterexamples

Datasets will be generated by using the Virtual simulator OpenLab ( https://openlab.app/ ). The student will study the datasets and decide upon a good setting to extract a finite state automaton for the drilling process. The student will also do a short survey of existing techniques for extracting finite state automata from process data. We present a novel algorithm that uses exact learning and abstraction to extract a deterministic finite automaton describing the state dynamics of a given trained RNN. We do this using Angluin's L*algorithm as a learner and the trained RNN as an oracle. Our technique efficiently extracts accurate automata from trained RNNs, even when the state vectors are large and require fine differentiation.arxiv.org

Scaling Laws for Language Models in Generative AI

Large Language Models (LLM) power today's most prominent language technologies in Generative AI like ChatGPT, which, in turn, are changing the way that people access information and solve tasks of many kinds.

A recent interest on scaling laws for LLMs has shown trends on understanding how well they perform in terms of factors like the how much training data is used, how powerful the models are, or how much computational cost is allocated. (See, for example, Kaplan et al. - "Scaling Laws for Neural Language Models”, 2020.)

In this project, the task will consider to study scaling laws for different language models and with respect with one or multiple modeling factors.

Advisor: Dario Garigliotti

Applications of causal inference methods to omics data

Many hard problems in machine learning are directly linked to causality [1]. The graphical causal inference framework developed by Judea Pearl can be traced back to pioneering work by Sewall Wright on path analysis in genetics and has inspired research in artificial intelligence (AI) [1].

The Michoel group has developed the open-source tool Findr [2] which provides efficient implementations of mediation and instrumental variable methods for applications to large sets of omics data (genomics, transcriptomics, etc.). Findr works well on a recent data set for yeast [3].

We encourage students to explore promising connections between the fiels of causal inference and machine learning. Feel free to contact us to discuss projects related to causal inference. Possible topics include: a) improving methods based on structural causal models, b) evaluating causal inference methods on data for model organisms, c) comparing methods based on causal models and neural network approaches.

References:

1. Schölkopf B, Causality for Machine Learning, arXiv (2019):  https://arxiv.org/abs/1911.10500

2. Wang L and Michoel T. Efficient and accurate causal inference with hidden confounders from genome-transcriptome variation data. PLoS Computational Biology 13:e1005703 (2017).  https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1005703

3. Ludl A and and Michoel T. Comparison between instrumental variable and mediation-based methods for reconstructing causal gene networks in yeast. arXiv:2010.07417  https://arxiv.org/abs/2010.07417

Advisors: Adriaan Ludl ,  Tom Michoel

Space-Time Linkage of Fish Distribution to Environmental Conditions

Conditions in the marine environment, such as, temperature and currents, influence the spatial distribution and migration patterns of marine species. Hence, understanding the link between environmental factors and fish behavior is crucial in predicting, e.g., how fish populations may respond to climate change.   Deriving this link is challenging because it requires analysis of two types of datasets (i) large environmental (currents, temperature) datasets that vary in space and time, and (ii) sparse and sporadic spatial observations of fish populations.

Project goal   

The primary goal of the project is to develop a methodology that helps predict how spatial distribution of two fish stocks (capelin and mackerel) change in response to variability in the physical marine environment (ocean currents and temperature).  The information can also be used to optimize data collection by minimizing time spent in spatial sampling of the populations.

The project will focus on the use of machine learning and/or causal inference algorithms.  As a first step, we use synthetic (fish and environmental) data from analytic models that couple the two data sources.  Because the ‘truth’ is known, we can judge the efficiency and error margins of the methodologies. We then apply the methodologies to real world (empirical) observations.

Advisors:  Tom Michoel , Sam Subbey . 

Towards precision medicine for cancer patient stratification

On average, a drug or a treatment is effective in only about half of patients who take it. This means patients need to try several until they find one that is effective at the cost of side effects associated with every treatment. The ultimate goal of precision medicine is to provide a treatment best suited for every individual. Sequencing technologies have now made genomics data available in abundance to be used towards this goal.

In this project we will specifically focus on cancer. Most cancer patients get a particular treatment based on the cancer type and the stage, though different individuals will react differently to a treatment. It is now well established that genetic mutations cause cancer growth and spreading and importantly, these mutations are different in individual patients. The aim of this project is use genomic data allow to better stratification of cancer patients, to predict the treatment most likely to work. Specifically, the project will use machine learning approach to integrate genomic data and build a classifier for stratification of cancer patients.

Advisor: Anagha Joshi

Unraveling gene regulation from single cell data

Multi-cellularity is achieved by precise control of gene expression during development and differentiation and aberrations of this process leads to disease. A key regulatory process in gene regulation is at the transcriptional level where epigenetic and transcriptional regulators control the spatial and temporal expression of the target genes in response to environmental, developmental, and physiological cues obtained from a signalling cascade. The rapid advances in sequencing technology has now made it feasible to study this process by understanding the genomewide patterns of diverse epigenetic and transcription factors as well as at a single cell level.

Single cell RNA sequencing is highly important, particularly in cancer as it allows exploration of heterogenous tumor sample, obstructing therapeutic targeting which leads to poor survival. Despite huge clinical relevance and potential, analysis of single cell RNA-seq data is challenging. In this project, we will develop strategies to infer gene regulatory networks using network inference approaches (both supervised and un-supervised). It will be primarily tested on the single cell datasets in the context of cancer.

Developing a Stress Granule Classifier

To carry out the multitude of functions 'expected' from a human cell, the cell employs a strategy of division of labour, whereby sub-cellular organelles carry out distinct functions. Thus we traditionally understand organelles as distinct units defined both functionally and physically with a distinct shape and size range. More recently a new class of organelles have been discovered that are assembled and dissolved on demand and are composed of liquid droplets or 'granules'. Granules show many properties characteristic of liquids, such as flow and wetting, but they can also assume many shapes and indeed also fluctuate in shape. One such liquid organelle is a stress granule (SG). 

Stress granules are pro-survival organelles that assemble in response to cellular stress and important in cancer and neurodegenerative diseases like Alzheimer's. They are liquid or gel-like and can assume varying sizes and shapes depending on their cellular composition. 

In a given experiment we are able to image the entire cell over a time series of 1000 frames; from which we extract a rough estimation of the size and shape of each granule. Our current method is susceptible to noise and a granule may be falsely rejected if the boundary is drawn poorly in a small majority of frames. Ideally, we would also like to identify potentially interesting features, such as voids, in the accepted granules.

We are interested in applying a machine learning approach to develop a descriptor for a 'classic' granule and furthermore classify them into different functional groups based on disease status of the cell. This method would be applied across thousands of granules imaged from control and disease cells. We are a multi-disciplinary group consisting of biologists, computational scientists and physicists. 

Advisors: Sushma Grellscheid , Carl Jones

Machine Learning based Hyperheuristic algorithm

Develop a Machine Learning based Hyper-heuristic algorithm to solve a pickup and delivery problem. A hyper-heuristic is a heuristics that choose heuristics automatically. Hyper-heuristic seeks to automate the process of selecting, combining, generating or adapting several simpler heuristics to efficiently solve computational search problems [Handbook of Metaheuristics]. There might be multiple heuristics for solving a problem. Heuristics have their own strength and weakness. In this project, we want to use machine-learning techniques to learn the strength and weakness of each heuristic while we are using them in an iterative search for finding high quality solutions and then use them intelligently for the rest of the search. Once a new information is gathered during the search the hyper-heuristic algorithm automatically adjusts the heuristics.

Advisor: Ahmad Hemmati

Machine learning for solving satisfiability problems and applications in cryptanalysis

Advisor: Igor Semaev

Hybrid modeling approaches for well drilling with Sintef

Several topics are available.

"Flow models" are first-principles models simulating the flow, temperature and pressure in a well being drilled. Our project is exploring "hybrid approaches" where these models are combined with machine learning models that either learn from time series data from flow model runs or from real-world measurements during drilling. The goal is to better detect drilling problems such as hole cleaning, make more accurate predictions and correctly learn from and interpret real-word data.

The "surrogate model" refers to  a ML model which learns to mimic the flow model by learning from the model inputs and outputs. Use cases for surrogate models include model predictions where speed is favoured over accuracy and exploration of parameter space.

Surrogate models with active Learning

While it is possible to produce a nearly unlimited amount of training data by running the flow model, the surrogate model may still perform poorly if it lacks training data in the part of the parameter space it operates in or if it "forgets" areas of the parameter space by being fed too much data from a narrow range of parameters.

The goal of this thesis is to build a surrogate model (with any architecture) for some restricted parameter range and implement an active learning approach where the ML requests more model runs from the flow model in the parts of the parameter space where it is needed the most. The end result should be a surrogate model that is quick and performs acceptably well over the whole defined parameter range.

Surrogate models trained via adversarial learning

How best to train surrogate models from runs of the flow model is an open question. This master thesis would use the adversarial learning approach to build a surrogate model which to its "adversary" becomes indistinguishable from the output of an actual flow model run.

GPU-based Surrogate models for parameter search

While CPU speed largely stalled 20 years ago in terms of working frequency on single cores, multi-core CPUs and especially GPUs took off and delivered increases in computational power by parallelizing computations.

Modern machine learning such as deep learning takes advantage this boom in computing power by running on GPUs.

The SINTEF flow models in contrast, are software programs that runs on a CPU and does not happen to utilize multi-core CPU functionality. The model runs advance time-step by time-step and each time step relies on the results from the previous time step. The flow models are therefore fundamentally sequential and not well suited to massive parallelization.

It is however of interest to run different model runs in parallel, to explore parameter spaces. The use cases for this includes model calibration, problem detection and hypothesis generation and testing.

The task of this thesis is to implement an ML-based surrogate model in such a way that many surrogate model outputs can be produced at the same time using a single GPU. This will likely entail some trade off with model size and maybe some coding tricks.

Uncertainty estimates of hybrid predictions (Lots of room for creativity, might need to steer it more, needs good background literature)

When using predictions from a ML model trained on time series data, it is useful to know if it's accurate or should be trusted. The student is challenged to develop hybrid approaches that incorporates estimates of uncertainty. Components could include reporting variance from ML ensembles trained on a diversity of time series data, implementation of conformal predictions, analysis of training data parameter ranges vs current input, etc. The output should be a "traffic light signal" roughly indicating the accuracy of the predictions.

Transfer learning approaches

We're assuming an ML model is to be used for time series prediction

It is possible to train an ML on a wide range of scenarios in the flow models, but we expect that to perform well, the model also needs to see model runs representative of the type of well and drilling operation it will be used in. In this thesis the student implements a transfer learning approach, where the model is trained on general model runs and fine-tuned on a most representative data set.

(Bonus1: implementing one-shot learning, Bonus2: Using real-world data in the fine-tuning stage)

ML capable of reframing situations

When a human oversees an operation like well drilling, she has a mental model of the situation and new data such as pressure readings from the well is interpreted in light of this model. This is referred to as "framing" and is the normal mode of work. However, when a problem occurs, it becomes harder to reconcile the data with the mental model. The human then goes into "reframing", building a new mental model that includes the ongoing problem. This can be seen as a process of hypothesis generation and testing.

A computer model however, lacks re-framing. A flow model will keep making predictions under the assumption of no problems and a separate alarm system will use the deviation between the model predictions and reality to raise an alarm. This is in a sense how all alarm systems work, but it means that the human must discard the computer model as a tool at the same time as she's handling a crisis.

The student is given access to a flow model and a surrogate model which can learn from model runs both with and without hole cleaning and is challenged to develop a hybrid approach where the ML+flow model continuously performs hypothesis generation and testing and is able to "switch" into predictions of  a hole cleaning problem and different remediations of this.

Advisor: Philippe Nivlet at Sintef together with advisor from UiB

Explainable AI at Equinor

In the project Machine Teaching for XAI (see  https://xai.w.uib.no ) a master thesis in collaboration between UiB and Equinor.

Advisor: One of Pekka Parviainen/Jan Arne Telle/Emmanuel Arrighi + Bjarte Johansen from Equinor.

Explainable AI at Eviny

In the project Machine Teaching for XAI (see  https://xai.w.uib.no ) a master thesis in collaboration between UiB and Eviny.

Advisor: One of Pekka Parviainen/Jan Arne Telle/Emmanuel Arrighi + Kristian Flikka from Eviny.

If you want to suggest your own topic, please contact Pekka Parviainen ,  Fabio Massimo Zennaro or Nello Blaser .

Theses on the topic of quantum computing

Available Topics

Currently, all thesis topics are already assigned. New theses will be announced by mid-2024.

Already assigned topics

Stochastic gradient line bayesian optimization (sglbo) for quantum computing (b.sc.).

Current algorithms for quantum computers are based on so-called variational approaches. In this process, quantum gates - the programmable building blocks of quantum computers - are parameterized with variables and numerically optimized using conventional computers. To do this, gradient-based methods are often applied. However, noise poses a particular challenge when it comes to optimization due to the probabilistic nature of quantum physics, as well as the significant measurement noise of today’s error-prone hardware. Optimization methods for quantum algorithms must therefore be able to deal with noise.

This is where machine learning optimization methods come into play. A highly promising optimization approach is Stochastic Gradient Line Bayesian Optimization (SGLBO). This uses a machine learning method (Bayesian Optimization) to control the optimization step by step. A recent publication demonstrated how this can give quantum algorithms an advantage over other optimization methods. This bachelor thesis will examine to what extent quantum neural networks can be optimized using the SGLBO method. Quantum neural networks function like artificial neural networks, but they are run on a quantum computer. In the first part of the thesis, the SGLBO method will be implemented and tested in Python. This will be followed by a comparison with other previously-implemented optimization methods under different noise influences.  Finally, the thesis will assess how well the optimization works on IBM's real quantum computing hardware. The thesis provides an exciting opportunity to address current optimization challenges using quantum computing and to make an important contribution in this field. Prior knowledge of numerical optimization is a great advantage, as well as a general interest in the topics of quantum computing and machine learning.

Quantum Kernel Methods for Solving Differential Equations (B.Sc, M.Sc)

In the DEGRAD-EL3-Q project, we are investigating how quantum computing methods can be used to analyze the lifetime of electrolyzers. The project is part of the lead project H2Giga and aims at advancing the industrial manufacturing process of electrolyzers. The mathematical description of how electrolyzers behave in operation can be modeled by differential equations. In this project, we want to explore the extent to which quantum kernel methods can be used to solve differential equations. In addition, a systematic comparison will be made with the quantum neural networks also studied in the project. This is an exciting and forward-looking topic in the superposition state of quantum computing and hydrogen research.

Evaluating the Use of Quantum Graph Neural Networks to Predict Molecular Properties (M.Sc.)

Scalable and cost-effective solutions for storing renewable energy are essential if we are to meet the world's increasing energy demand and simultaneously mitigate climate change. The conversion of electricity to hydrogen, as well as the reverse combustion process, can play an important role. To make catalysis processes in hydrogen production efficient, new materials are constantly being studied. Machine learning methods are already being used to simulate and calculate catalysis properties. Graph-based neural networks (GNN) are proving to be particularly promising in this respect. Since the prediction of potential surfaces and other relevant properties takes place at molecular and atomic level, the use of quantum computers is also being considered. First approaches to implement GNNs on quantum computers have already been published. The objective of the master thesis is to determine the suitability of quantum GNNs for predicting molecular properties. To this end, depending on prior knowledge, an understanding of GNNs, as well as some basic knowledge of quantum computing, must first be acquired. In-depth knowledge of electrocatalysis is not necessarily required. Towards the end of the master thesis, the developed approaches can be tested and evaluated on a real quantum computer.