Reinforcement Learning Architectures (1992)
| Venue: | Proceedings ISKIT'92 International Symposium on Neural Information Processing |
| Citations: | 4 - 0 self |
BibTeX
@INPROCEEDINGS{Sutton92reinforcementlearning,
author = {Richard S. Sutton},
title = {Reinforcement Learning Architectures},
booktitle = {Proceedings ISKIT'92 International Symposium on Neural Information Processing},
year = {1992}
}
Years of Citing Articles
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
Reinforcement learning is the learning of a mapping from situations to actions so as to maximize a scalar reward or reinforcement signal. The learner is not told which action to take, as in most forms of learning, but instead must discover which actions yield the highest reward by trying them. In the most interesting and challenging cases, actions affect not only the immediate reward, but also the next situation, and through that all subsequent rewards. These two characteristics -- trial-and-error search and delayed reward -- are the two most important distinguishing features of reinforcement learning. In this paper I present a brief overview of the development of reinforcement learning architectures over the past decade, including reinforcement-comparison, actor-critic, and Q-learning architectures. Finally, I present Dyna, a class of architectures based on reinforcement learning but which go beyond trial-and-error learning to include a learned internal model of the world. By ...
