BET: Explaining Deep Reinforcement Learning through The Error-Prone Decisions

TL;DR

This paper introduces BET, a novel interpretable model that identifies error-prone states in deep reinforcement learning agents by analyzing decision consistency and state neighborhoods, improving explanation fidelity especially in complex environments.

Contribution

BET is a new self-interpretable structure that pinpoints error-prone states in DRL by modeling state neighborhoods and decision uniformity, advancing explainability in complex scenarios.

Findings

BET outperforms existing models in explanation fidelity.

BET effectively identifies error-prone states in various RL environments.

First to explain complex multi-agent scenarios like StarCraft II transparently.

Abstract

Despite the impressive capabilities of Deep Reinforcement Learning (DRL) agents in many challenging scenarios, their black-box decision-making process significantly limits their deployment in safety-sensitive domains. Several previous self-interpretable works focus on revealing the critical states of the agent's decision. However, they cannot pinpoint the error-prone states. To address this issue, we propose a novel self-interpretable structure, named Backbone Extract Tree (BET), to better explain the agent's behavior by identify the error-prone states. At a high level, BET hypothesizes that states in which the agent consistently executes uniform decisions exhibit a reduced propensity for errors. To effectively model this phenomenon, BET expresses these states within neighborhoods, each defined by a curated set of representative states. Therefore, states positioned at a greater distance from these representative benchmarks are more prone to error. We evaluate BET in various popular RL environments and show its superiority over existing self-interpretable models in terms of explanation fidelity. Furthermore, we demonstrate a use case for providing explanations for the agents in StarCraft II, a sophisticated multi-agent cooperative game. To the best of our knowledge, we are the first to explain such a complex scenarios using a fully transparent structure.