mlsanityData to 14 Oct 2026

Command-Space Counterfactual Explanations for Pareto-Conditioned Reinforcement Learning

arXiv 2608.14963 ↗First posted 15 Aug 2026cs.LG cs.AI cs.HCcited by 0

On the map: Deep reinforcement learning methods → Interpretable and logic-guided RL

Abstract (arXiv)

Pareto Conditioned Networks learn multiple multi-objective reinforcement learning behaviours by conditioning a single policy on a desired return command. However, the local mapping from command and state to action remains opaque. We propose command-space counterfactual explanations for PCNs: given a fixed state, original command, and foil action, we search, in a black-box setting, for a minimally changed desired-return command under which the same trained policy would choose the foil. Our contributions are threefold. First, we formulate PCN explanations as return-command interventions, using a return-only PCN variant that avoids the added ambiguity of horizon-conditioning. Second, we adapt adversarial machine learning methods to reinforcement-learning explanations. Third, we introduce a boundary-seeded directional search that improves over purely local optimization in the command-action landscape, resulting in our proposed approach CF-ZOO. The resulting explanations are actionable and intuitively expressed in the user's own preferences: "If your trade-off had shifted slightly towards X, the agent would have chosen Y."

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Interpretable and logic-guided RL, one of 15 directions in Deep reinforcement learning methods. Builds on 23 of its 32 references that are in the map. First posted on arXiv 15 Aug 2026.

Builds on

23 of 32 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024SAFE-RL: Saliency-Aware Counterfactual Explainer for Deep Reinforcement Learning Policies28 Apr
  2. 2024ACTER: Diverse and Actionable Counterfactual Sequences for Explaining and Diagnosing RL Policies9 Feb

Other references · 21

  1. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
  2. 2017ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Modelscited by 666
  3. 2019Explaining Machine Learning Classifiers through Diverse Counterfactual Explanationscited by 614
  4. 2021A Practical Guide to Multi-Objective Reinforcement Learning and Planningcited by 256
  5. 2014A Survey of Multi-Objective Sequential Decision-Makingcited by 242
  6. 2018Adversarial Robustness Toolbox v1.0.0cited by 215
  7. 2019A Generalized Algorithm for Multi-Objective Reinforcement Learning and Policy Adaptationcited by 177
  8. 2020Counterfactual Explanations and Algorithmic Recourses for Machine Learning: A Reviewcited by 174
  9. 2020Multi-Objective Counterfactual Explanationscited by 157
  10. 2018Query-Efficient Hard-label Black-box Attack:An Optimization-based Approachcited by 130
  11. 2018Dynamic Weights in Multi-Objective Deep Reinforcement Learningcited by 100
  12. 2022PD-MORL: Preference-Driven Multi-Objective Reinforcement Learning Algorithmcited by 58
9 more
  1. 2021Counterfactual State Explanations for Reinforcement Learning Agents via Generative Deep Learningcited by 39
  2. 2021Counterfactual Explanations in Sequential Decision Making Under Uncertaintycited by 37
  3. 2022Pareto Conditioned Networkscited by 25
  4. 2021Explainable Reinforcement Learning for Broad-XAI: A Conceptual Framework and Surveycited by 24
  5. 2023GANterfactual-RL: Understanding Reinforcement Learning Agents' Strategies through Visual Counterfactual Explanationscited by 17
  6. 2023Explaining Reinforcement Learning Agents Through Counterfactual Action Outcomescited by 14
  7. 2022Redefining Counterfactual Explanations for Reinforcement Learning: Overview, Challenges and Opportunitiescited by 13
  8. 2023RACCER: Towards Reachable and Certain Counterfactual Explanations for Reinforcement Learningcited by 6
  9. 2020Adversarial Attacks on Deep Algorithmic Trading Policiescited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Counterfactual Explanations for Continuous Action Reinforcement Learning
  2. 2025STACHE: Local Black-Box Explanations for Reinforcement Learning Policies
  3. 2026Explaining Reinforcement Learning Decisions in Self-adaptive Systems
  4. 2026Counterfactual Rollouts for Offline Contrastive Goal-Conditioned Reinforcement Learning
  5. 2024Explaining Reinforcement Learning: A Counterfactual Shapley Values Approach
  6. 2026Reward-Conditioned Reinforcement Learning
  7. 2022Towards Interpretable Deep Reinforcement Learning Models via Inverse Reinforcement Learning
  8. 2021PsiPhi-Learning: Reinforcement Learning with Demonstrations using Successor Features and Inverse Temporal Difference Learning