Cognitive Science > Reinforcement Learning

[edit] Introduction

How can systems learn to achieve certain goals, solely from information on how well they do? If you tell a system that it has made mistakes but you do not tell it what exactly these mistakes were and when exactly the mistakes happened, can the system improve? This is the basic problem of Reinforcement Learning.

[edit] Sections

1. The Reinforcement Learning Problem
2. Formalization
3. Learning in Known Environments
4. Learning in Unknown Environments
5. Learning in Continuous Environments