Research group logo

Teaching

Bachelor & Master Thesis Guidelines

Apply for a thesis

Send your topic of interest, Transcript of Records, and CV to patrick.knab@tu-clausthal.de.

Please allow 2-3 weeks for a response after your initial request.

We offer thesis topics, subject to availability, across several active research areas. The information below consolidates the current public-facing process, expectations, and submission guidance.

Available Thesis Topics

Adversarial Attacks & Model Robustness

Understanding failure modes of ML systems and improving robustness under distribution shift and attack.

AI for Software Engineering

Automated requirements engineering, AI-assisted GUI prototyping, and adaptive user interfaces that evolve with users, contexts, and changing requirements.

Mathematical Theory behind ML

Theoretical perspectives on machine learning models, optimization, and generalization behaviour.

Generative AI for Building Information Modeling (BIM)

Using generative AI to interact with, generate, and evaluate BIM models in digital planning workflows.

Tabular Data / Deep Learning for Tabular Data / Tabular Foundation Models

Learning robust representations, designing strong deep learning architectures, and developing next-generation foundation models for structured tabular data.

Combining Transduction and Induction in Programming by Example

Using Programming by Example as a testbed to combine instance-specific transductive reasoning with more general inductive methods.

Concept Based Deep Learning & Computer Vision in the Wild

Interpretable deep learning, concept bottlenecks, and robust visual reasoning outside narrowly curated benchmarks.

Process

  1. Step 1. Submit your application with your topic of interest, Transcript of Records, and CV to patrick.knab@tu-clausthal.de.
  2. Step 2. If a topic is available, define a concrete direction together with the supervising researcher.
  3. Step 3. Write and submit an exposé using the provided proposal template.
  4. Step 4. Receive approval for the exposé.
  5. Step 5. Register the thesis independently with the Studienbüro.
  • What the Expose Should Cover
    • Use the provided proposal template; maximum 2 pages excluding references.
    • Include your name and a preliminary title at the top.
    • Introduction and background together should stay compact and should not exceed one page.
    • Goals and work plan should clearly state research questions, planned approach, evaluation strategy, expected outcomes, and realistic milestones.
    • Distinguish must-have goals from nice-to-have goals.
    • Write scientifically: frame the work as an investigation rather than a guaranteed improvement claim.
    • Your first task is typically a review of relevant literature and tools.
    • Avoid formulas, algorithms, and low-level technical details; prefer high-level, intuitive descriptions.
  • Report Basics
    • Language: English
    • Template: official thesis LaTeX template provided by your supervisor
    • Target length: Bachelor 30-50 pages, Master 50-70 pages, excluding references and appendices
  • Working With Your Advisor
    • You are responsible for scheduling all meetings.
    • Meet with your advisor at least once per month.
    • Plan a halfway-point progress check with your advisor.
    • Come prepared with slides and results that demonstrate meaningful progress.
    • Send your slides to your supervisor before each meeting.
    • Failing to schedule and prepare meetings can negatively affect your grade.
  • Report Expectations
    • Start writing early and keep detailed notes throughout the project, including experiment settings.
    • Discuss the report outline with your advisor before writing the full draft.
    • Include an abstract.
    • Write for readers with a computer science background; avoid repeating textbook material.
    • Be concise and move long code listings or implementation details into an appendix if needed.
    • Use examples and figures throughout the report.
    • Recommended reading: Writing for Computer Science by Justin Zobel.
  • Defense Format
    • The final presentation is scheduled after the thesis submission.
    • After the presentation, supervisors ask 3-5 comprehension questions.
    • You may consult your slides or thesis for up to 30 seconds before answering each question.
  • What Must Be Handed In
    • For binding requirements and deadlines, consult your Studienordnung or contact the Prüfungsamt.
    • Submit source code, data, and setup instructions via an online repository such as GitHub, or optionally on CD-ROM / USB stick.
    • If the full dataset is too large for physical media, include a representative sample.
  • Use of Generative AI
    • Students bear full responsibility for all submitted content and must be able to explain and defend it.
    • The group does not police how students use generative AI tools.
    • Hallucinated references are treated as a quality issue and can negatively affect the grade.
    • Repeated hallucinations, or hallucinations in critical parts of the thesis, can lead to a substantially lower grade or failure.
  • Assessment

    The thesis report remains the primary basis for assessment, together with the submitted artifacts, such as the source code or demonstrator. In addition, supervisors prepare comprehension questions for the final presentation to assess the student’s understanding of the topic. The final presentation is also evaluated based on its overall quality, the presentation of the thesis work, and the student’s ability to answer questions.