Reinforcement Learning


Overview


Introduction TD-Learning Algorithms Applet Follow Up Source Code References


  Group Members

Tim Eden 2229705 s2229705@cse.unsw.edu.au
Anthony Knittel 2232071 aek@cse.unsw.edu.au
Raphael van Uffelen 3041561 raphis@gmx.de


  Overview

The aim of this tutorial is to give a student with some understanding of Artificial Intelligence methods an in-depth look at Reinforcement Learning, one particular approach to Machine Learning. The information is divided up into a number of sections.

Firstly, there is an Introduction to Reinforcement Learning. Following the introduction is an explanation of TD-Learning, and how it relates to Reinforcement Learning. Two particular Algorithms, Q-Learning and Sarsa will then be explained, along with an example to illustrate their differences. To demonstrate these algorithms at work, there is an Applet which learns to play the game of "Cat and Mouse". There is an interactive tutorial to go with the applet which helps explain how a reinforcement based learner operates.

The Follow Up section is a discussion of how our code implemements the algorithms. Also, the complete Source Code for the applet is freely available to be viewed or downloaded. A list of References that were used in the making of these web pages has been compiled, including links to on-line sources where available.

To get started, click on the link below:

Start Here