Challenges:

Major challenges include learning from limited training data, ensuring broad experimental applicability, and avoiding introducing bias into the analysis.

Methods:

Your task shall be to develop a cutting-edge human-AI collaboration workflow. This involvest raining AI models on small expert-annotated datasets, using explainable AI to uncover biases, and leveraging active learning for iterative refinement with human feedback. You shall also explore self-supervised learning to uncover hidden patterns, enabling potentially completely new insights into complex scientific data.

Interdisciplinarity:

This Masters Project is an interdisciplinary collaboration of the Robust Data Science Group (Prof. Muma) with the Laboratory Astrophysics Group (Prof. Kuschel).

Prerequisites:

• Strong foundation in machine learning and AI: Essential for developing and refiningthe complex algorithms required.
• Programming skills: Proficiency in Python is crucial for implementing the pipeline efficiently.
• Motivation and Interest: A high level of enthusiasm for developing and applying advanced methods and working with real physics data is key to success in this project.

How to Apply?

Please send an E-Mail to michael.muma@tu-darmstadt.de with your CV and transcript.