An Introduction to Reinforcement Learning

12:15 PM - 12:40 PM on July 17, 2016, Room CR4

Jessica Forde

Audience level:


Reinforcement learning is a subfield of machine learning focused on discovering ‘optimal policies’: robust strategies to achieve a desired objective under varying states of the world. In this talk, we will provide an overview of terminology in reinforcement learning and sample Python programs for basic algorithms to learn policies. We will also discuss the state of the art in reinforcement learning and the ways in which reinforcement learning can be applied to real world problems.


Reinforcement learning has gained media attention in recent years for the creation of software to solve games such as Space Invaders and Go. The basics of reinforcement learning, however, can be performed using simple probability and linear algebra. Using scientific computing libraries in Python such as numpy and pandas, we will explore basic algorithms in reinforcement learning. We will also discuss more contemporary reinforcement learning research and the relationship between deep learning and reinforcement learning.