Entity · paper

Explainable Reinforcement Learning via Physics-Aware Policy Distillation

paperactiveprovisionalexplainable-reinforcement-learning-via-physics-aware-policy-distillation-80f250da·1 events·first seen 2d ago

Aliases: Explainable Reinforcement Learning via Physics-Aware Policy Distillation

Co-occurring entities

Twin Delayed DDPG

More like this (12)

Physics-EnhAnced Reinforcement Learning Pass the Baton: Trajectory-Relayed On-Policy Distillation The Physics of Multi-Turn Long-Horizon Planning: From Pre-training to Post-training via Single- and Multi-Teacher On-Policy Agentic Distillation Routing-based On-Policy Distillation on-policy distillation Weak-to-Strong Generalization via Direct On-Policy Distillation Explainable Reinforcement Learning for assisting Air Traffic Controllers Dense Supervision, Sparse Updates: On the Sparsity and Geometry of On-Policy Distillation Learning from the Self-future: On-policy Self-distillation for dLLMs Geometric Action Model for Robot Policy Learning On-Policy Distillation for LLM Safety: A Routing Approach to Template-Robust Realignment Entropy-Regularized Reinforcement Learning

Recent events (1)

3arXiv · cs.LG·2d ago·source ↗

Physics-aware policy distillation makes deep RL agents interpretable via decision tree surrogates

A new arXiv preprint presents a policy distillation framework that transfers a high-performance TD3 reinforcement learning agent into a shallow decision tree surrogate for interpretable continuous control, demonstrated on the Inverted Pendulum benchmark. The method uses physics-aware features and 'Noisy Oracle Rollouts' to match teacher performance while providing global and local interpretability. A key finding is a fundamental trade-off: discretizing control induces Bang-Bang actuation and a bimodal limit cycle, though BIBO stability is preserved. The work targets safety-critical deployment contexts where regulatory compliance and human-agent trust require explainability.

AI Safety Research Explainable Reinforcement Learning via Physics-Aware Policy Distillation Twin Delayed DDPG