Interpret Policies in Deep Reinforcement Learning using SILVER with RL-Guided Labeling: A Model-level Approach to High-dimensional and Multi-action Environments

2025-10-22

AI Summary

Deep reinforcement learning (DRL) policies suffer from poor interpretability, hindering human trust; existing SILVER frameworks are restricted to low-dimensional, binary-action settings. Method: We propose an RL-guided annotation mechanism that integrates policy outputs into boundary point identification, extending SILVER for the first time to high-dimensional, multi-action environments. Our approach combines SHAP-based feature attribution, compact feature extraction, and lightweight surrogate models (e.g., decision trees) to achieve model-level, behaviorally consistent explanations. Results: Evaluated on the Atari benchmark, our method preserves original policy performance while significantly enhancing explanation transparency. Human-subject studies demonstrate superior clarity and trustworthiness compared to baselines. The core contribution is the first interpretable DRL framework overcoming SILVER’s limitations in state-action dimensionality, enabling faithful, scalable explainability for high-dimensional, multi-action DRL policies.

Technology Category

• Machine Learning: Transparent, Interpretable, Explainable ML
• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Deep reinforcement learning (RL) achieves remarkable performance but lacks interpretability, limiting trust in policy behavior. The existing SILVER framework (Li, Siddique, and Cao 2025) explains RL policy via Shapley-based regression but remains restricted to low-dimensional, binary-action domains. We propose SILVER with RL-guided labeling, an enhanced variant that extends SILVER to multi-action and high-dimensional environments by incorporating the RL policy's own action outputs into the boundary points identification. Our method first extracts compact feature representations from image observations, performs SHAP-based feature attribution, and then employs RL-guided labeling to generate behaviorally consistent boundary datasets. Surrogate models, such as decision trees and regression-based functions, are subsequently trained to interpret RL policy's decision structure. We evaluate the proposed framework on two Atari environments using three deep RL algorithms and conduct human-subject study to assess the clarity and trustworthiness of the derived interpretable policy. Results show that our approach maintains competitive task performance while substantially improving transparency and human understanding of agent behavior. This work advances explainable RL by transforming SILVER into a scalable and behavior-aware framework for interpreting deep RL agents in high-dimensional, multi-action settings.

Problem

• Extends interpretable RL to high-dimensional multi-action environments
• Explains deep reinforcement learning policies using surrogate models
• Improves policy transparency while maintaining competitive task performance

Innovation

• Extracts compact feature representations from images
• Uses RL-guided labeling for boundary datasets
• Trains surrogate models to interpret policy decisions