This practical research course explores advanced machine learning methods to empower autonomous agents with intelligent decision-making capabilities. Students will delve into:

·       Generative Models for Decision Making

·       Reinforcement Learning (RL)

·       Imitation Learning

·       Multi-Agent Systems

·       Uncertainty Quantification

·       Learning Prediction Models of Physical Processes

·       Time-Series Modeling

·       Discovery and Inference of Latent Variables

Each student will choose one of the offered topics, implement one or several algorithms, and evaluate them against available baselines using standard benchmark tasks. The course emphasizes hands-on experimentation, requiring students to document their findings in a detailed report. Students will work in teams of two, closely collaborating with their supervisor with the aim of achieving publishable results. This course provides students with their first experience in running a research project in machine learning, including algorithm design, evaluation, benchmarking, deploying algorithms on HPC hardware, and paper writing.

-        The discussed algorithms have to be implemented successfully.

-        The experiments need to be conducted scientifically and need to be well documented.

-        The final report is well written and well structured

-        The final presentation is well prepared

 Recommendations:

-        Experience in Machine Learning is recommended.

-        Python experience is recommended

-        We will use the PyTorch deep learning library. Some prior knowledge in this is helpful but             not necessary.