Seminar

Joint seminar on deep learning for molecular representations, property prediction, and de-novo molecular design — with a focus on geometric deep learning.

Recent advances in Graph Neural Networks and Geometric Deep Learning, with applications to drug discovery, brain analysis, and social networks.

Seminar on deep reinforcement learning, the predecessor of the 2022/23 edition.

A computer-science perspective on Monte Carlo: variance reduction, MCMC, SGD, rare event simulation, evolutionary algorithms.

Dynamical processes on networks: brain networks, social networks, epidemics — analyzed via diffusion models, stochastic simulation and deep learning.

Foundations of reinforcement learning combined with a practical group project on the game Connect 4.