Introduction to Online and Reinforcement Learning

3 Credits

Review of probability and random processes, basic reinforcement learning framework, learning from streaming data, actions in response to changing environment through Markov Decision Processes, elements of artificial intelligence. Exploration-Exploitation tradeoffs through bandit problems, and different methods for reinforcement learning including dynamic programming, Monte Carlo methods, temporal difference and Q-learning. Approximate solutions for very large state space systems, policy iteration and actor critic methods, introduction of deep reinforcement learning. Credit will not be given for both ECE 340 and ECE 440 . Repeat Status: Course may be repeated. Prerequisites: MATH 231 or MATH 309