CS 4789 - Introduction to Reinforcement Learning

Spring. 3 credits. Letter grades only.

Prerequisite: CS 3780 or equivalent. Co-meets with CS 5789.

S. Dean.

Reinforcement Learning is one of the most popular paradigms for modelling interactive learning and sequential decision making in dynamical environments. This course introduces the basics of Reinforcement Learning and the Markov Decision Process. The course will cover algorithms for planning and learning in Markov Decision Processes. We will discuss potential applications of Reinforcement Learning and their implications. We will study and implement classic Reinforcement Learning algorithms.

Outcome 1: Identify the differences between Reinforcement Learning and traditional Supervised Learning and grasp the key definitions of Markov Decision Processes.

Outcome 2: Analyze the performance of the class planning algorithms and learning algorithms for Markov Decision Process.

Outcome 3: Implement classic algorithms and demonstrate their performance on benchmarks.