Saved in:
Bibliographic Details
Main Authors: Malloy, Tailia, Klinger, Tim, Liu, Miao, Riemer, Matthew, Tesauro, Gerald, Sims, Chris R.
Format: Preprint
Published: 2020
Subjects:
Online Access:https://arxiv.org/abs/2011.11517
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909611017109504
author Malloy, Tailia
Klinger, Tim
Liu, Miao
Riemer, Matthew
Tesauro, Gerald
Sims, Chris R.
author_facet Malloy, Tailia
Klinger, Tim
Liu, Miao
Riemer, Matthew
Tesauro, Gerald
Sims, Chris R.
contents This paper introduces an information-theoretic constraint on learned policy complexity in the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) reinforcement learning algorithm. Previous research with a related approach in continuous control experiments suggests that this method favors learning policies that are more robust to changing environment dynamics. The multi-agent game setting naturally requires this type of robustness, as other agents' policies change throughout learning, introducing a nonstationary environment. For this reason, recent methods in continual learning are compared to our approach, termed Capacity-Limited MADDPG. Results from experimentation in multi-agent cooperative and competitive tasks demonstrate that the capacity-limited approach is a good candidate for improving learning performance in these environments.
format Preprint
id arxiv_https___arxiv_org_abs_2011_11517
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Consolidation via Policy Information Regularization in Deep RL for Multi-Agent Games
Malloy, Tailia
Klinger, Tim
Liu, Miao
Riemer, Matthew
Tesauro, Gerald
Sims, Chris R.
Artificial Intelligence
This paper introduces an information-theoretic constraint on learned policy complexity in the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) reinforcement learning algorithm. Previous research with a related approach in continuous control experiments suggests that this method favors learning policies that are more robust to changing environment dynamics. The multi-agent game setting naturally requires this type of robustness, as other agents' policies change throughout learning, introducing a nonstationary environment. For this reason, recent methods in continual learning are compared to our approach, termed Capacity-Limited MADDPG. Results from experimentation in multi-agent cooperative and competitive tasks demonstrate that the capacity-limited approach is a good candidate for improving learning performance in these environments.
title Consolidation via Policy Information Regularization in Deep RL for Multi-Agent Games
topic Artificial Intelligence
url https://arxiv.org/abs/2011.11517