Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning

TL;DR

This paper introduces a causal influence-based reward mechanism for multi-agent deep reinforcement learning, promoting coordination and communication by incentivizing agents to have a significant impact on others' actions, leading to improved social behaviors.

Contribution

It presents a novel decentralized influence reward method using counterfactual reasoning, enabling agents to learn diverse policies and emergent communication without centralized training.

Findings

Enhanced coordination and communication in social dilemma environments.

Dramatic improvements in learning speed and policy diversity.

Agents develop meaningful communication protocols.

Abstract

We propose a unified mechanism for achieving coordination and communication in Multi-Agent Reinforcement Learning (MARL), through rewarding agents for having causal influence over other agents' actions. Causal influence is assessed using counterfactual reasoning. At each timestep, an agent simulates alternate actions that it could have taken, and computes their effect on the behavior of other agents. Actions that lead to bigger changes in other agents' behavior are considered influential and are rewarded. We show that this is equivalent to rewarding agents for having high mutual information between their actions. Empirical results demonstrate that influence leads to enhanced coordination and communication in challenging social dilemma environments, dramatically increasing the learning curves of the deep RL agents, and leading to more meaningful learned communication protocols. The influence rewards for all agents can be computed in a decentralized way by enabling agents to learn a model of other agents using deep neural networks. In contrast, key previous works on emergent communication in the MARL setting were unable to learn diverse policies in a decentralized manner and had to resort to centralized training. Consequently, the influence reward opens up a window of new opportunities for research in this area.