Statistical Physics for Optimization & Learning

This lecture introduces the concepts of statistical physics applied to optimization and learning problems. The instructor presents tools such as the replica method, cavity method, and message passing algorithms. Students will work on open research problems in groups, applying the theoretical tools learned in the lecture. Topics covered include graph coloring, low-rank matrix factorization, and generalized linear regression. The lecture also explores the application of statistical physics in recommendation systems and learning rules using neural networks. Evaluation is based on weekly homework and contributions to solving open problems.