Deep Reinforcement Learning

Catalog description

Deep Reinforcement Learning (DRL) studies how agents can learn complex behaviors through trial-and-error interaction with an environment, combining reinforcement learning principles with deep neural networks for high-dimensional perception and control. This course provides a rigorous, hands-on introduction to modern DRL methods, from foundational Markov decision processes and value-based learning to policy gradients, actor–critic algorithms, and scalable techniques used in real-world decision systems. Students will implement core algorithms, analyze stability and sample-efficiency challenges, and apply DRL to benchmark control tasks while developing intuition for when and why specific methods work.