Reward Functions for Accelerated Learning
Abstract
This paper discusses why traditional reinforcement learning methods, and algorithms applied to those models, result in poor performance in situated domains characterized by multiple goals, noisy state, and inconsistent reinforcement. We propose a methodology for designing reinforcement functions that take advantage of implicit domain knowledge in order to accelerate learning in such domains. The methodology involves the use of heterogeneous reinforcement functions and progress estimators, and applies to learning in domains with a single agent or with multiple agents. The methodology is experimentally validated on a group of mobile robots learning a foraging task.
Cite
Text
Mataric. "Reward Functions for Accelerated Learning." International Conference on Machine Learning, 1994. doi:10.1016/B978-1-55860-335-6.50030-1Markdown
[Mataric. "Reward Functions for Accelerated Learning." International Conference on Machine Learning, 1994.](https://mlanthology.org/icml/1994/mataric1994icml-reward/) doi:10.1016/B978-1-55860-335-6.50030-1BibTeX
@inproceedings{mataric1994icml-reward,
title = {{Reward Functions for Accelerated Learning}},
author = {Mataric, Maja J.},
booktitle = {International Conference on Machine Learning},
year = {1994},
pages = {181-189},
doi = {10.1016/B978-1-55860-335-6.50030-1},
url = {https://mlanthology.org/icml/1994/mataric1994icml-reward/}
}