Publication - Multi-Agent Learning with Policy Prediction
| Authors: | Zhang, Chongjie; Lesser, Victor | ||||
| Title: | Multi-Agent Learning with Policy Prediction | ||||
| Abstract: | Due to the non-stationary environment, learning in multi-agent systems is a challenging problem. This paper first introduces a new gradient-based algorithm by using policy prediction in basic gradient ascent. We prove that this modification results in a stronger notion of convergence than basic gradient ascent, that is, strategies converge to a Nash equilibrium within a restricted class of iterated games. Motivated by this modification, we then propose a new practical multi-agent reinforcement learning (MARL) algorithm exploiting approximate policy prediction. Empirical results show that it converges faster and in a wider variety of situations than state-of-the-art MARL algorithms. | ||||
| Keywords: | Learning, Multi-Agent Systems | ||||
| Publication: | Proceedings of the 24th AAAI Conference on Artificial Intelligence, pp. 927 - 934 | ||||
| Location: | Atlanta | ||||
| Date: | 2010 | ||||
| Sources: |
PDF: http://www.cs.umass.edu/~chongjie/pub/aaai2010_zhang.pdf PDF: /Documents/cjzhang/aaai10_zhang.pdf |
||||
| Reference: | Zhang, Chongjie; Lesser, Victor. Multi-Agent Learning with Policy Prediction. Proceedings of the 24th AAAI Conference on Artificial Intelligence, pp. 927-934. 2010. | ||||
| bibtex: | @inproceedings{Zhang-487,
author = "Chongjie Zhang and Victor Lesser",
title = "{Multi-Agent Learning with Policy Prediction}",
booktitle = "Proceedings of the 24th AAAI Conference on
Artificial Intelligence",
pages = "927-934",
year = "2010",
address = "Atlanta",
url = "http://mas.cs.umass.edu/paper/487",
}
|
||||
