AILGMAJul 24, 2020

Value-Decomposition Multi-Agent Actor-Critics

arXiv:2007.12306v4141 citations
AI Analysis

This addresses training efficiency and performance trade-offs in multi-agent reinforcement learning for domains like game AI, but it is incremental as it builds on existing value-decomposition and actor-critic methods.

The paper tackles the incompatibility between QMIX and A2C in multi-agent reinforcement learning by proposing value-decomposition actor-critics (VDACs), which improve median performance on StarCraft II micromanagement tasks over other actor-critic methods.

The exploitation of extra state information has been an active research area in multi-agent reinforcement learning (MARL). QMIX represents the joint action-value using a non-negative function approximator and achieves the best performance, by far, on multi-agent benchmarks, StarCraft II micromanagement tasks. However, our experiments show that, in some cases, QMIX is incompatible with A2C, a training paradigm that promotes algorithm training efficiency. To obtain a reasonable trade-off between training efficiency and algorithm performance, we extend value-decomposition to actor-critics that are compatible with A2C and propose a novel actor-critic framework, value-decomposition actor-critics (VDACs). We evaluate VDACs on the testbed of StarCraft II micromanagement tasks and demonstrate that the proposed framework improves median performance over other actor-critic methods. Furthermore, we use a set of ablation experiments to identify the key factors that contribute to the performance of VDACs.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes