From Few to More: Large-scale Dynamic Multiagent Curriculum Learning
This addresses the problem of complex dynamics and state-action space explosion in large-scale multiagent systems for researchers in reinforcement learning, though it appears incremental as it builds on existing curriculum learning ideas.
The paper tackles the challenge of learning in large-scale multiagent systems by proposing a dynamic curriculum learning approach that starts with few agents and increases their number, using a novel network structure to handle varying input sizes, resulting in greatly improved performance over state-of-the-art methods.
A lot of efforts have been devoted to investigating how agents can learn effectively and achieve coordination in multiagent systems. However, it is still challenging in large-scale multiagent settings due to the complex dynamics between the environment and agents and the explosion of state-action space. In this paper, we design a novel Dynamic Multiagent Curriculum Learning (DyMA-CL) to solve large-scale problems by starting from learning on a multiagent scenario with a small size and progressively increasing the number of agents. We propose three transfer mechanisms across curricula to accelerate the learning process. Moreover, due to the fact that the state dimension varies across curricula,, and existing network structures cannot be applied in such a transfer setting since their network input sizes are fixed. Therefore, we design a novel network structure called Dynamic Agent-number Network (DyAN) to handle the dynamic size of the network input. Experimental results show that DyMA-CL using DyAN greatly improves the performance of large-scale multiagent learning compared with state-of-the-art deep reinforcement learning approaches. We also investigate the influence of three transfer mechanisms across curricula through extensive simulations.