Learning Progress Driven Multi-Agent Curriculum

TL;DR

This paper introduces a novel curriculum learning method for multi-agent reinforcement learning that uses TD-error based learning progress to better control task difficulty, leading to improved performance on challenging benchmarks.

Contribution

It proposes a new curriculum control approach based on learning progress, addressing variance and credit assignment issues in MARL.

Findings

Outperforms state-of-the-art baselines on three benchmarks.

Uses TD-error based learning progress for curriculum control.

Alleviates issues of high variance and credit assignment difficulty.

Abstract

The number of agents can be an effective curriculum variable for controlling the difficulty of multi-agent reinforcement learning (MARL) tasks. Existing work typically uses manually defined curricula such as linear schemes. We identify two potential flaws while applying existing reward-based automatic curriculum learning methods in MARL: (1) The expected episode return used to measure task difficulty has high variance; (2) Credit assignment difficulty can be exacerbated in tasks where increasing the number of agents yields higher returns which is common in many MARL tasks. To address these issues, we propose to control the curriculum by using a TD-error based *learning progress* measure and by letting the curriculum proceed from an initial context distribution to the final task specific one. Since our approach maintains a distribution over the number of agents and measures learning progress rather than absolute performance, which often increases with the number of agents, we alleviate problem (2). Moreover, the learning progress measure naturally alleviates problem (1) by aggregating returns. In three challenging sparse-reward MARL benchmarks, our approach outperforms state-of-the-art baselines.