ECE 586: Markov Decision Processes and Reinforcement Learning (Spring 2019)

Course Information

  • Instructor: Dimitrios Katselis, Email: katselis@illinois.edu

  • TA: Joseph Lubars, Email: lubars2@illinois.edu

  • Schedule: M W 11-12:20, ECEB 3081

  • Office hours : F 10:30-11:45, ECEB 3042 (Dimitris), W 2:30-3:30, ECEB 2036 (Joseph)

Outline

  • Markov Chains

  • Gradient Descent, Stochastic Gradient Descent

  • Neural Networks

  • Multi-Armed Bandits

  • Markov Decision Processes

  • Dynamic Programming

  • Numerical Methods: Value and Policy Iteration

  • Monotone policies

  • Q-Learning

  • Stochastic Approximation

  • ODE Method

  • Policy Gradient