Policy Optimization with Sparse Global Contrastive Explanations
Not on the map. Why?
Abstract (arXiv)
We develop a Reinforcement Learning (RL) framework for improving an existing behavior policy via sparse, user-interpretable changes. Our goal is to make minimal changes while gaining as much benefit as possible. We define a minimal change as having a sparse, global contrastive explanation between the original and proposed policy. We improve the current policy with the constraint of keeping that global contrastive explanation short. We demonstrate our framework with a discrete MDP and a continuous 2D navigation domain.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 23 of its 33 references that are in the map. First posted on arXiv 13 Jul 2022.
Builds on
23 of 33 references.
References · 23
- 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,400
- 2017Explanation in Artificial Intelligence: Insights from the Social Sciencescited by 1,201
- 2017Counterfactual Explanations without Opening the Black Box: Automated Decisions and the GDPRcited by 1,086
- 2017Dynamic Routing Between Capsulescited by 689
- 2021Interpretable Machine Learning: Fundamental Principles and 10 Grand Challengescited by 364
- 2018Actionable Recourse in Linear Classificationcited by 327
- 2016Data-Efficient Off-Policy Policy Evaluation for Reinforcement Learningcited by 285
- 2018Techniques for Interpretable Machine Learningcited by 237
- 2019Model-Agnostic Counterfactual Explanations for Consequential Decisionscited by 204
- 2018Programmatically Interpretable Reinforcement Learningcited by 192
- 2017Interpretable Convolutional Neural Networkscited by 161
- 2015Visualizing and Understanding Recurrent Networkscited by 158
11 more
- 2020Explainable Reinforcement Learning: A Surveycited by 139
- 2018Boolean Decision Rules via Column Generationcited by 79
- 2021Contrastive Explanations for Model Interpretabilitycited by 75
- 2018Toward Interpretable Deep Reinforcement Learning with Linear Model U-Treescited by 65
- 2018Contrastive Explanations for Reinforcement Learning in terms of Expected Consequencescited by 44
- 2021Bootstrapping Fitted Q-Evaluation for Off-Policy Inferencecited by 40
- 2018Contrastive Explanations with Local Foil Treescited by 34
- 2019Generating Counterfactual and Contrastive Explanations using SHAPcited by 23
- 2020What Did You Think Would Happen? Explaining Agent Behaviour Through Intended Outcomescited by 20
- 2019Optimization Methods for Interpretable Differentiable Decision Trees in Reinforcement Learningcited by 5
- 2020Tradeoff-Focused Contrastive Explanation for MDP Planningcited by 5
Built on by
Citing papers per year: 2022: 1; 2023: 0; 2024: 0; 2025: 1.
2 papers.