Research map / Label-efficient and robust learning / Recognition and long-tail
Joint Shapley values: a measure of joint feature importance
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Measure coalitions by their average marginal contribution, rather than interaction relative to constituent features | Feature attribution · Majority game (n=3) | value for the full coalition: Shapley interaction index: -2; Shapley-Taylor, k=3: -2; Joint Shapley, k=3: \nicefrac{3}{21} | Mixed |
| Use presence-adjusted rather than absolute local values for global attribution | Joint Shapley global attribution for binary features · Binary-feature ML attribution | consistency with local intuitions | Helped |
| Include correlated features x1 and x2 together in the coalition | Joint Shapley · Simulated correlated features, f2(x), x2=1-x1 | global joint Shapley value: 0 | No clear effect |
| Include correlated features x1 and x2 together in the coalition | Joint Shapley · Simulated correlated features, f3(x), x2=1-x1 | global joint Shapley value: 0.122 -> 0.098 | Helped |
| Increase explanation order from k=2 to k=3 | Joint Shapley · Boston housing | global joint Shapley value for LSTAT: 0.63 -> 0.42 | Hurt |
Closest papers by meaning
- Shapley Residuals: Quantifying the limits of the Shapley value for explanations (NeurIPS 2021 · similarity 0.92)
- shapiq: Shapley Interactions for Machine Learning (NeurIPS 2024 · similarity 0.90)
- RankSHAP: Shapley Value Based Feature Attributions for Learning to Rank (ICLR 2025 · similarity 0.88)
- Unlocking the Game: Estimating Games in Möbius Representation for Explanation and High-Order Interaction Detection (AAAI 2025 · similarity 0.87)
- KernelSHAP-IQ: Weighted Least Square Optimization for Shapley Interactions (ICML 2024 · similarity 0.87)
- A Unified Framework for Provably Efficient Algorithms to Estimate Shapley Values (NeurIPS 2025 · similarity 0.86)
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.