Distilling Deep Reinforcement Learning into Interpretable Fuzzy Rules: An Explainable AI Framework

Sanup S. Araballi; Simon Khan; Chilukuri K. Mohan

cs.AIFeb 24, 2026

Distilling Deep Reinforcement Learning into Interpretable Fuzzy Rules: An Explainable AI Framework

Authors: Sanup S. Araballi, Simon Khan, Chilukuri K. Mohan

Organizations: Dept. of EECS, Syracuse University, NY 13244, USA · Air Force Research Laboratory, Rome, NY 13441, USA

Abstract

Deep Reinforcement Learning (DRL) agents achieve remarkable performance in continuous control but remain opaque, hindering deployment in safety-critical domains. Existing explainability methods either provide only local insights (SHAP, LIME) or employ over-simplified surrogates failing to capture continuous dynamics (decision trees). This work proposes a Hierarchical Takagi-Sugeno-Kang (TSK) Fuzzy Classifier System (FCS) distilling neural policies into human-readable IF-THEN rules through K-Means clustering for state partitioning and Ridge Regression for local action inference. Three quantifiable metrics are introduced: Fuzzy Rule Activation Density (FRAD) measuring explanation focus, Fuzzy Set Coverage (FSC) validating vocabulary completeness, and Action Space Granularity (ASG) assessing control mode diversity. Dynamic Time Warping (DTW) validates temporal behavioral fidelity. Empirical evaluation on Lunar Lander(Continuous) shows the Triangular membership function variant achieves 81.48% $\pm$ 0.43% fidelity, outperforming Decision Trees by 21 percentage points. The framework exhibits statistically superior interpretability (FRAD = 0.814 vs. 0.723 for Gaussian, $p < 0.001$ ) with low MSE (0.0053) and DTW distance (1.05). Extracted rules such as ``IF lander drifting left at high altitude THEN apply upward thrust with rightward correction'' enable human verification, establishing a pathway toward trustworthy autonomous systems.

Figures & tables

Figure 1: Hierarchical TSK FCS Architecture. Level 1 partitions state space via K-Means clustering into operational regions. Level 2 learns local TSK consequents within each region via Ridge Regression. Final actions obtained through normalized weighted aggregation of active local models.

Model	Fidelity	MSE	DTW	FRAD	FSC
	(%)
Simple-MLP	96.84	0.0016	0.55	N/A	N/A
	$\pm$ 1.80	$\pm$ 0.001	$\pm$ 0.20
FCS-Tri.	81.48	0.0053	1.05	0.814	0.933
	$\pm$ 0.43	$\pm$ 0.001	$\pm$ 0.17	$\pm$ 0.009	$\pm$ 0.002
FCS-Gaus.	81.38	0.0037	0.87	0.723	0.974

Table 1: Performance Comparison of Surrogate Models. Higher is better for Fidelity and FRAD; lower is better for MSE and DTW.

#Rules	Fidelity (%)	MSE	FRAD	FSC
4	97.83 $\pm$ 0.5	0.00069	0.863	0.937
8	95.83 $\pm$ 0.7	0.00119	0.801	0.963
16	81.48 $\pm$ 0.4	0.00534	0.814	0.933

Table 2: Impact of Rule Count on FCS-Triangular Performance.

Figure 2: Trajectory alignment: PPO teacher (black solid) vs. FCS surrogate (red dashed) over 50 timesteps. Close tracking demonstrates high behavioral fidelity (DTW = 1.03).

Figure 3: Learned Guassian membership functions for Position $X$ . K-Means partitioned the dimension into overlapping, locally concentrated regions. Compact support ensures localized activations.

Figure 4: Rule activation for sample state. Only Rules 2 and 16 are significantly activated ( $\alpha_{2}\approx 0.7$ , $\alpha_{16}\approx 0.28$ ), yielding FRAD = 0.89.

Figure 5: Fidelity vs. rule count. The 4-rule model achieves peak (97.8%), validating the “Less is More” principle. Beyond 4 rules, over-partitioning degrades performance.

Interpretability in RL

Interpretability in reinforcement learning makes learned policies, decisions, and internal representations understandable and auditable.

See all 71 papers

Explainable Reinforcement Learning via Physics-Aware Policy Distillation
Jul 27, 2026Shaker Al-Tamari, Waled KadourRL Control Interpretability in RL
Dept. of Electrical and Computer Engineering RPTU Kaiserslautern-Landau Kaiserslautern, Germany
ORCAID: Oblique Rule-Based Continuous-Action Interpretation for Deep RL Policies
Jul 8, 2026Ignacio D. Lopez-Miguel, Ezio Bartocci, Thomas Eiter +1Reinforcement Learning Interpretable ML
TU Wien Vienna, Austria
Explainably Safe Reinforcement Learning
Jun 3, 2026Sabine Rieder, Stefan Pranger, Debraj Chakraborty +2Reinforcement Learning Interpretability in RL
Masaryk University · Technical University of Munich · Graz University of Technology

Jul 27, 2026J/K move · Enter open · S save

Explainable Reinforcement Learning via Physics-Aware Policy Distillation

Shaker Al-Tamari, Waled Kadour

Dept. of Electrical and Computer Engineering RPTU Kaiserslautern-Landau Kaiserslautern, Germany

RL Control Interpretability in RL Policy Distillation

Open paper AskPDF arXiv

In safety-critical sectors such as robotics and automotive engineering, the deployment of Deep Reinforcement Learning (DRL) is often hindered by the black-box nature of deep neural networks. This lack of transparency poses significant challenges for regulatory compliance and human-agent trust. This paper presents an experimental study aimed at making high-performance continuous control DRL systems interpretable. A policy distillation framework is implemented using the classic Inverted Pendulum benchmark. A high-performance Twin Delayed DDPG (TD3) agent serves as an opaque, continuous teacher model, whose policy is distilled into an interpretable student surrogate based on a shallow Decision Tree. By leveraging a custom physics-aware feature and "Noisy Oracle Rollouts" for dataset generation, the distillation process achieves performance equivalent to the expert teacher. Furthermore, comparative control theory analysis reveals a fundamental trade-off: transitioning from continuous to discrete rule-based control induces high-frequency Bang-Bang actuation and a stable bimodal limit cycle. Simulation results indicate that Bounded-Input Bounded-Output (BIBO) stability is maintained while providing both global and local interpretability for safe autonomous systems.

Abstract

Interpretability in RL

Explore similar work

Explainable Reinforcement Learning via Physics-Aware Policy Distillation

ORCAID: Oblique Rule-Based Continuous-Action Interpretation for Deep RL Policies

Explainably Safe Reinforcement Learning