Temporally Interpretable Differentiable Decision Trees

Eisuke Hirota; Aarav Sane; Rohan Paleja

cs.LGOct 7, 2026

Temporally Interpretable Differentiable Decision Trees

Authors: Eisuke Hirota, Aarav Sane, Rohan Paleja

Organizations: Department of Computer Science, Purdue University · Thomas Lord Department of Computer Science, University of Southern California

Abstract

Interpretability offers a solution to safe autonomy by providing transparency into an agent's underlying decision-making model. Within sequential-decision making tasks, differentiable decision trees (DDTs) are one approach to such interpretability, maintaining automatic-differentiable policies while providing humans with a discrete tree-based visualization. Nonetheless, current implementations of DDTs are not well-suited for sequential-decision making domains, as there exists an inherent mismatch between a tree's single-timestep behavior and a human's multi-timestep planning. Our work thus introduces time as a new dimension of interpretability, coined as temporal interpretability, and demonstrates how temporal abstractions via action chunking improve it. We achieve this by first introducing two novel policy gradient algorithms that incorporate action chunking. Additionally, to maintain parameter-efficient trees, we develop an information-theoretic tree restructuring algorithm that modifies the tree during training. Across four simulation environments, we find that warm-starting action chunked DDTs from a distilled action chunked policy is the most effective way to obtain temporally interpretable trees: they match neural network policies in three of the four domains while using up to 80 $\%$ fewer parameters. Our code is available at https://github.com/ei5uke/temp-interp.

Figures & tables

Figure 1 : Overview figure. (a) Temporal Interpretability. We provide an example depicting how humans increase in confidence when provided with an agent’s short-term plan and temporal decision-making (an autonomous vehicle’s plan is visualized with blue arrows). (b) Temporally Abstracted Policy Gradients. We introduce two novel policy gradient methods that enable action chunked DDTs to be temporally interpretable. (c) Information-Theoretic Tree Restructuring. Introducing action chunking to trees increases parameter count, hence we dynamically deepen or prune leaves to ensure the model only contains sub-branches we need most.

	Environments
Algorithms	LK	IP	LL	LL-H
MLP	$185.3\pm 2.9$ 497	$1000.0\pm 0.0$ 369	$293.8\pm 2.3$ 450	$272.1\pm 7.2$ $450$
MLP-Ensemble	$190.2\pm 2.0$ 650	$1000.0\pm 0.0$ 522	$296.0\pm 3.2$ 756	$270\pm 3.2$ $756$
MLP-Prediction	$193.3\pm 0.9$ 650	$1000.0\pm 0.0$ 522	$294.6\pm 1.6$ 756	$273.2\pm 5.6$ $756$
MLP-Big	$191.4\pm 1.6$ $5057$	$1000.0\pm 0.0$ $4545$	$292.4\pm 3.8$ $4866$	$269.6\pm 4.0$ $4866$
MLP-Big-Ensemble	$189.2\pm 3.8$ $5642$	$1000.0\pm 0.0$ $5130$	$293.3\pm 270$ $6036$	$274.3\pm 3.2$ $6036$

Table 1 : Performance comparison of all algorithms in the four simulation domains, across three runs in an evaluation environment separate from the training environment. We report two values in each cell: the top value depicts the episodic return (mean

\pm

s.e.) and the bottom value depicts the number of parameters (constant values for MLPs and mean

\pm

s.e. for DDTs). For CART, we count a feature index and a threshold per decision node and the weights and bias of each leaf’s linear model.

Panel of Figure 2 — Figure 2 : Figures depicting the trend in the number of leaves within DDTs when using our deepening / pruning algorithm for each simulation environment. Each curve showcases the mean and s.e. for a specific environment across three seeds, depicted in the dark curve and shaded regions respectively. Note that we run IP and LK for less timesteps, hence leading to the curves cutting off earlier.

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Algorithm	LK	IP	LL	LL-H
DDT-Ensemble NoLeaves	$18\pm 0.0$	$10\pm 0.0$	$14\pm 0.0$	$20.0\pm 4.9$
DDT-Pred NoLeaves	$25.3\pm 6.0$	$10\pm 0.0$	$14\pm 0.0$	$26.0\pm 4.9$
Warm-Ensemble NoLeaves	$18\pm 0.0$	$10\pm 0.0$	$14\pm 0.0$	$26.0\pm 4.9$
Warm-Pred NoLeaves	$18\pm 0.0$	$10\pm 0.0$	$14\pm 0.0$	$33.3\pm 9.4$

Appendix

Table 3 : Number of parameters used only for the logic flow of DDTs across the four domains and three runs (mean

\pm

s.e.).

Panel of Figure 5 — Figure 5 : Figures depicting the trend in the performance against the EMA coefficient in each simulation environment. The black curve depicts the regression over the results across all seeds.

Parameter	Value
RPO
Actor learning rate	$5\textrm{e}{-4}$
Critic learning rate	$5\textrm{e}{-4}$
PPO clip rate	$0.2$
RPO uniform bonus ( Rahman et al., 2025 )	$0.5$
MLP

Appendix

Table 4 : Shared hyperparameters used for all environments.

	Domain
Parameter	LK	IP	LL	LL-H
DDT
Starting $\#$ of leaves	$2$	$2$	$2$	$4$
Maximum $\#$ of CART leaves	$2$	$2$	$2$	$4$
Information-Theoretic Tree Restructuring
Minimum steps since last restructure $n$	$200$	$600$	$1000$	$1000$

Appendix

Table 5 : Unique hyperparameters used for each environment for ITTR.

Explore similar work

Interpretable reinforcement learning with decision-tree pruning
Aug 7, 2026Mark Leon Ringer, Michel TokicStructured Pruning
Faculty of Mathematics, Informatics and Statistics, Ludwig-Maximilians-University Munich, Munich, Germany · Siemens AG, Data & Artificial Intelligence, Otto-Hahn-Ring 6, 81739 Munich, Germany
dtControl2+ $\varepsilon$ : Trading Optimality for Explainability in MDPs via Decision Trees
Jul 28, 2026Tereza Kinská, Jan Křetínský, Tobias Meggendorfer +2Decision Trees Markov Decision Processes
Masaryk University, Brno, Czech Republic · Technical University of Munich, Munich, Germany · Lancaster University Leipzig, Leipzig, Germany +1
SPOTting the Future: Lookahead Explanations for Deep Reinforcement Learning
Jul 28, 2026Tamar Gozlan, Claudia V. GoldmanLookahead Traffic Signal Control
Benin School of Computer Science and Engineering, The Hebrew University of Jerusalem · Hebrew University Business School, Data Science Department, The Hebrew University of Jerusalem

Aug 7, 2026, cs.LGJ/K move · Enter open · S save

Interpretable reinforcement learning with decision-tree pruning

Mark Leon Ringer, Michel Tokic

Faculty of Mathematics, Informatics and Statistics, Ludwig-Maximilians-University Munich, Munich, Germany · Siemens AG, Data & Artificial Intelligence, Otto-Hahn-Ring 6, 81739 Munich, Germany

Structured Pruning

Open paper AskPDF arXiv

Reinforcement learning policies are difficult to inspect, but interpreting them is a prerequisite for trustworthiness. Converting a trained policy into explicit decision-tree rules improves transparency and the resulting artifacts often remain too complex for human understanding. We present a pruning process that simplifies such rule-based policies while preserving task performance and making edits to the policy auditable. The process defines a small set of structural and usage-aware operators and evaluates candidate edits by re-executing the policy to measure return and interpretability proxies. This exposes an transformation process from complex to compact policy structures. We investigate this approach on classic control and MuJoCo benchmarks, where pruning traces reveal consistent interpretability improvements while maintaining high performance.

Abstract

Explore similar work

Interpretable reinforcement learning with decision-tree pruning

dtControl2+ε\varepsilonε: Trading Optimality for Explainability in MDPs via Decision Trees

SPOTting the Future: Lookahead Explanations for Deep Reinforcement Learning

dtControl2+ $\varepsilon$ : Trading Optimality for Explainability in MDPs via Decision Trees