Research map / Learning theory and algorithms / Prediction and decision losses
Explaining Preferences with Shapley Values
Robert Hu, Siu Lun Chau, Jaime Ferrando Huertas, Dino Sejdinovic
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 GPM instead of UPM | Preference prediction · Synthetic | Test AUC: 0.71_\pm0.01 -> 0.98_{\pm0.00} | Improved |
| Use GPM instead of UPM | Preference prediction · Chameleon | Test AUC: 0.80_{\pm0.07} -> 0.92_{\pm0.07} | Improved |
| Use GPM instead of UPM | Preference prediction · Pokémon | Test AUC: 0.82_{\pm0.00} -> 0.86_{\pm0.00} | Improved |
| Use context-specific C-GPM instead of UPM | Preference prediction · Tennis | Test AUC: 0.52_{\pm0.02} -> 0.58_{\pm0.02} | Improved |
| Use context-specific C-GPM instead of UPM | Preference prediction · Website | Test AUC: 0.65_{\pm0.01} -> 0.66_{\pm0.01} | No clear effect |
| Explain pairwise preferences with Pref-SHAP rather than applying SHAP to UPM utilities | Synthetic pairwise preference explanations · Synthetic | recovery of the designed feature mechanism | Improved |
| Explain pairwise preferences with Pref-SHAP rather than applying SHAP to UPM utilities | Pokémon battle explanations · Pokémon | identification of relevant preference features | Improved |
| Explain pairwise preferences with Pref-SHAP rather than applying SHAP to UPM utilities | Chameleon contest explanations · Chameleon | consistency of jaw-length effect | Improved |
| Concatenate both items' features and apply ordinary SHAP instead of using the shared-feature preferential value function | Pairwise preference explanations · Synthetic | interpretability of feature attributions | Hurt |
Closest papers by meaning
- shapiq: Shapley Interactions for Machine Learning (NeurIPS 2024 · similarity 0.87)
- Explaining Probabilistic Models with Distributional Values (ICML 2024 · similarity 0.85)
- RankSHAP: Shapley Value Based Feature Attributions for Learning to Rank (ICLR 2025 · similarity 0.85)
- Provably Accurate Shapley Value Estimation via Leverage Score Sampling (ICLR 2025 · similarity 0.85)
- InstaSHAP: Interpretable Additive Models Explain Shapley Values Instantly (ICLR 2025 · similarity 0.85)
- DU-Shapley: A Shapley Value Proxy for Efficient Dataset Valuation (NeurIPS 2024 · similarity 0.84)
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.
