Machine Learning for Systems and Control

The course will lay the methodological groundwork for machine learning in signals, systems, and control contexts and corresponding application domains. The course covers three modules: supervised learning, unsupervised learning, and reinforcement learning. Throughout the course, students will delve into topics such as linear regression, logistic regression, classification, neural networks, Bayesian supervised learning, and learn about model assessment techniques. In addition, the course covers clustering and matrix factorization, as well as fundamentals of reinforcement learning and its connection with dynamic programming. Furthermore, students will gain insight into model-free RL control algorithms like Sarsa and Q-learning, and approximation methods in value space.