Compositional Concept-Based Neuron-Level Interpretability for Deep Reinforcement Learning

TL;DR

This paper introduces a neuron-level interpretability method for deep reinforcement learning models, using concept-based explanations to improve transparency and align neural activations with human-understandable concepts.

Contribution

It presents a novel approach to interpret DRL networks by formalizing atomic concepts and analyzing neuron activations, enhancing understanding of internal decision mechanisms.

Findings

Effectively identifies meaningful concepts in DRL models

Aligns neuron activations with human-understandable concepts

Provides faithful explanations of network decision-making

Abstract

Deep reinforcement learning (DRL), through learning policies or values represented by neural networks, has successfully addressed many complex control problems. However, the neural networks introduced by DRL lack interpretability and transparency. Current DRL interpretability methods largely treat neural networks as black boxes, with few approaches delving into the internal mechanisms of policy/value networks. This limitation undermines trust in both the neural network models that represent policies and the explanations derived from them. In this work, we propose a novel concept-based interpretability method that provides fine-grained explanations of DRL models at the neuron level. Our method formalizes atomic concepts as binary functions over the state space and constructs complex concepts through logical operations. By analyzing the correspondence between neuron activations and concept functions, we establish interpretable explanations for individual neurons in policy/value networks. Experimental results on both continuous control tasks and discrete decision-making environments demonstrate that our method can effectively identify meaningful concepts that align with human understanding while faithfully reflecting the network's decision-making logic.