Research map / Learning theory and algorithms / Prediction and decision losses
Joint Distribution–Informed Shapley Values for Sparse Counterfactual Explanations
Lei You, Yijun Bian, Lele Cao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use counterfactual-informed Shapley with OT coupling instead of uniform coupling (ablation) | p-SHAP (COLA) · Hotel Bookings, 80% effect | number of modified features: 7.01 -> 2.50 | Improved |
| use OT coupling instead of random coupling (ablation) | p-SHAP (COLA) · Hotel Bookings, 80% effect | number of modified features: 10.63 -> 2.50 | Improved |
| use OT coupling instead of random coupling (ablation) | p-SHAP (COLA) · COMPAS, 80% effect | number of modified features: 2.58 -> 1.25 | Improved |
| use counterfactual information in the Shapley background instead of the random training-set baseline (with OT coupling) (ablation) | p-SHAP (COLA) · Hotel Bookings, 80% effect | number of modified features: 8.55 -> 2.50 | Improved |
| use counterfactual information in the Shapley background instead of random baseline (uniform coupling) (ablation) | p-SHAP (COLA) · Hotel Bookings, 80% effect | number of modified features: 7.10 -> 7.01 | Mixed |
| apply COLA with p-SHAP to refine counterfactuals | original counterfactuals · four datasets, 80% effect | share of original feature changes: 13%-25% of the feature changes | Improved |
| apply COLA with p-SHAP to reach full counterfactual effect | original counterfactuals · four datasets, 100% effect | share of original feature changes: 26%-45% of the feature changes | Improved |
| use OT coupling instead of known exact factual-counterfactual alignment (ablation) | p-SHAP (COLA) · German Credit MILP benchmark | modified features vs MILP optimum | Mixed |
Closest papers by meaning
- CounterFlowNet: From Minimal Changes to Meaningful Counterfactual Explanations (NeurIPS 2026 · similarity 0.88)
- Model Reconstruction Using Counterfactual Explanations: A Perspective From Polytope Theory (NeurIPS 2024 · similarity 0.88)
- Generating Likely Counterfactuals Using Sum-Product Networks (ICLR 2025 · similarity 0.88)
- Promoting Counterfactual Robustness through Diversity (AAAI 2024 · similarity 0.87)
- Faithful Model Explanations through Energy-Constrained Conformal Counterfactuals (AAAI 2024 · similarity 0.87)
- A New Approach to Backtracking Counterfactual Explanations: A Unified Causal Framework for Efficient Model Interpretability (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
