MMI706 REINFORCEMENT LEARNING

Course Objectives

Students who successfully complete this course are expected to acquire the knowledge of:

• the fundamentals of reinforcement learning
• Markov Decision Processes
• model-free prediction
• model-free control
• value function approximation
• policy gradient methods
• exploration and exploitation

At the end of the course, students will be able to pursue graduate studies in reinforcement learning related research areas.

Course Content

This course aims to give background knowledge on several topics related to reinforcement learning and provide an environment for practical applications. Multi-armed Bandits, Monte Carlo methods, Markov Decision Processes, Dynamic Programming and Temporal-Difference Learning are some of the core topics that will be covered through lectures. The course aims to balance theory and practice in that it will involve students implementing all of the described algorithms, testing those algorithms in different game environments, and reading recent research papers on the reinforcement learning field.

Course Learning Outcomes

1. Students will be familiar with reinforcement learning concepts,

2. Students will gain ability to apply recent techniques in reinforcement learning to different games,

3. Students will be able to implement and modify the algorithms in reinforcement learning,

4. Students will be able to learn the current trends in reinforcement learning, read related research papers and have a thorough understanding of the recent methodologies,

5. Students will gain the ability to creatively apply their current knowledge to produce new and original thoughts, ideas, processes in the field of reinforcement learning.