An Entropy Regularization Free Mechanism for Policy-based Reinforcement Learning
Abstract
Policy-based reinforcement learning methods suffer from the policy collapse problem. We find valued-based reinforcement learning methods with {\epsilon}-greedy mechanism are capable of enjoying three characteristics, Closed-form Diversity, Objective-invariant Exploration and Adaptive Trade-off, which help value-based methods avoid the policy collapse problem. However, there does not exist a parallel mechanism for policy-based methods that achieves all three characteristics. In this paper, we propose an entropy regularization free mechanism that is designed for policy-based methods, which achieves Closed-form Diversity, Objective-invariant Exploration and Adaptive Trade-off. Our experiments show that our mechanism is super sample-efficient for policy-based methods and boosts a policy-based baseline to a new State-Of-The-Art on Arcade Learning Environment.
Cite
@article{arxiv.2106.00707,
title = {An Entropy Regularization Free Mechanism for Policy-based Reinforcement Learning},
author = {Changnan Xiao and Haosen Shi and Jiajun Fan and Shihong Deng},
journal= {arXiv preprint arXiv:2106.00707},
year = {2021}
}
Comments
arXiv admin note: text overlap with arXiv:2105.03923