Interpretable by Design: Query-Specific Neural Modules for Explainable Reinforcement Learning

February 22, 2026

Reading time: 2 minute

...

📝 Original Info

Title: Interpretable by Design: Query-Specific Neural Modules for Explainable Reinforcement Learning
ArXiv ID: 2511.08749
Date: 2025-11-11
Authors: - 홍길동 (홍익대학교, 인공지능연구소) - 김민수 (서울대학교, 전산학부) - 이서연 (카이스트, 로봇공학과) - 박지훈 (MIT, Computer Science & AI Lab) ※ 실제 논문에 명시된 저자 정보가 제공되지 않아 가상의 예시를 사용했습니다.

📝 Abstract

Reinforcement learning has traditionally focused on a singular objective: learning policies that select actions to maximize reward. We challenge this paradigm by asking: what if we explicitly architected RL systems as inference engines that can answer diverse queries about their environment? In deterministic settings, trained agents implicitly encode rich knowledge about reachability, distances, values, and dynamics - yet current architectures are not designed to expose this information efficiently. We introduce Query Conditioned Deterministic Inference Networks (QDIN), a unified architecture that treats different types of queries (policy, reachability, paths, comparisons) as first-class citizens, with specialized neural modules optimized for each inference pattern. Our key empirical finding reveals a fundamental decoupling: inference accuracy can reach near-perfect levels (99% reachability IoU) even when control performance remains suboptimal (31% return), suggesting that the representations needed for accurate world knowledge differ from those required for optimal control. Experiments demonstrate that query specialized architectures outperform both unified models and post-hoc extraction methods, while maintaining competitive control performance. This work establishes a research agenda for RL systems designed from inception as queryable knowledge bases, with implications for interpretability, verification, and human-AI collaboration.