Master Thesis Projects at the Interface of Neuroscience and Machine Learning

The Max Planck Institute for Biological Cybernetics in Tübingen is renowned for its research on information processing in the brain. Using a multidisciplinary approach, our scientists employ experimental, theoretical, and computational methods to explore perception, memory, decision-making, motor performance, and more. We are equipped with state-of-the-art facilities and maintain close collaborations with neighboring Max Planck Institutes and the University of Tübingen. 

Master Thesis Projects at the Interface of Neuroscience and Machine Learning

The Neuroinformatics Research Group develops new methods for analyzing structural and functional MRI data of the human brain. We focus on two key aspects: the brain as a complex adaptive system with emergent patterns and the data-intensive nature of fMRI, enhanced by ultra-high field (UHF) MRI scanners. We currently offer the following projects:

1. Uncertainty-guided Unsupervised Domain Adaptation for Brain Segmentation at UHF MRI

• Develop an uncertainty estimation scheme for brain MRI segmentation to identify regions where the model is less confident.
• Build upon recently implemented UDA research frameworks, and use the uncertainty maps to guide the domain adaptation process at ultra-high field strength.

2. Reward Shaping in Reinforcement Learning for Enhanced Brain MRI Segmentation

• Design a reward function that incorporates expert feedback, penalizing inaccuracies and rewarding accurate segmentations.
• Use this shaped reward function to train RL agents to perform better segmentation tasks.

3. Improving Diagnosis of Alzheimer’s Disease Progression with Inherently Interpretable Models

• Evaluate inherently interpretable models for common important features of mild cognitive impairment to Alzheimer’s disease progression.
• Test the applicability of novel techniques, such as Kolmogorov-Arnold Networks, to identify these features while improving performance.

Your Profile:

• Enrollment in a relevant field of study such as computer science, biomedical engineering, cognitive science, neuroscience, or a comparable field
• Proficiency in Python programming including experience with libraries such as NumPy, SciPy, and Matplotlib
• Basic understanding of MRI data formats (e.g., .nii, .nii.gz, .img/.hdr) is an advantage but not required
• Ability to quickly learn new technical tools and concepts

The Max Planck Society is committed to employing more disabled people; applications from severely disabled people are therefore strongly encouraged. The Max Planck Society also wants to increase the number of women in areas where they are underrepresented. Women are therefore expressly encouraged to apply.

Interested?

Then send a short email outlining your previous experience and motivation to Julius Steiglechner (julius.steiglechner@tue.mpg.de)