About this role
Summer Semester 2027 – fixed term of 3–6 months Can explicit knowledge about objects make visual recognition more reliable? In this thesis, you will investigate how symbolic requirements can be combined with neural approaches in industrial applications. A neural model will extract visual concepts such as objects, parts, attributes, and spatial relationships. Structured knowledge will then be used either to verify predictions after inference or to guide the model during training.
YOUR TASKS
- Represent selected object knowledge as logical, relational, and geometric requirements.
- Investigate post-hoc verification and requirements-aware learning as two approaches to combining structured knowledge with neural predictions.
- Evaluate whether the resulting system improves prediction consistency, robustness, interpretability, or data efficiency compared with a purely neural baseline.
- The project will focus on a limited set of objects and requirements that can be evaluated reliably.
- The knowledge can be represented using classical rule sets or ontologies, while neural models identify the corresponding concepts in images.
- The expected outcome is a methodology and prototype for requirements-driven visual reasoning.
- You will investigate when symbolic knowledge helps detect or correct inconsistent predictions.
- A critical assessment of the underlying assumptions, limitations, and potential failure cases will be an essential part of the thesis.
YOUR PROFILE
- You are currently pursuing an advanced master's degree in Computer Science, Artificial Intelligence, Data Science, Robotics, Mathematics, Engineering, or a related field.
- You have a good foundation in machine learning and Python.
- Experience with computer vision, PyTorch, logic, knowledge graphs, semantic technologies, or neuro-symbolic AI is an advantage.
- You are interested in combining learned representations with explicit knowledge and critically evaluating the assumptions behind both approaches.
- You can quickly grasp complex technical challenges, work independently and systematically, and communicate your findings clearly within an interdisciplinary team.
- You have good written and spoken English skills.
You don't meet every requirement yet? We still encourage you to apply. We value curiosity, a strong willingness to learn, and the ability to approach complex problems systematically. You can also develop relevant skills during your thesis.
Contact: Sarah Disch