Prioritized Guidance for Efficient Multi-Agent Reinforcement Learning Exploration

Abstract

Exploration efficiency is a challenging problem in multi-agent reinforcement learning (MARL), as the policy learned by confederate MARL depends on the collaborative approach among multiple agents. Another important problem is the less informative reward restricts the learning speed of MARL compared with the informative label in supervised learning. In this work, we leverage on a novel communication method to guide MARL to accelerate exploration and propose a predictive network to forecast the reward of current state-action pair and use the guidance learned by the predictive network to modify the reward function. An improved prioritized experience replay is employed to better take advantage of the different knowledge learned by different agents which utilizes Time-difference (TD) error more effectively. Experimental results demonstrates that the proposed algorithm outperforms existing methods in cooperative multi-agent environments. We remark that this algorithm can be extended to supervised learning to speed up its training.

arXiv commentTheequations (7)-(10) in the paper are incorrectly derived, and need to be withdrawn and revised in many places

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Continuous control with deep reinforcement learning
• Trust Region Policy Optimization
• Prioritized Experience Replay
• Dueling Network Architectures for Deep Reinforcement Learning
• Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments
• Deep Reinforcement Learning: An Overview
• Actor-Mimic: Deep Multitask and Transfer Reinforcement Learning
• Toward Game Level Generation from Gameplay Videos
• Learning latent state representation for speeding up exploration