Research map / Learning theory and algorithms / Prediction and decision losses
Sampling Permutations for Shapley Value Estimation
Rory Mitchell, Joshua Cooper, Eibe Frank, Geoffrey Holmes
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 herding instead of MC-antithetic | permutation sampling · Discrepancy d=10, n=1000 | discrepancy: 0.027 -> 0.013 | Improved |
| use SBQ instead of MC-antithetic | permutation sampling · Discrepancy d=10, n=100 | discrepancy: 0.084 -> 0.056 | Improved |
| use herding instead of MC-antithetic | permutation sampling · Discrepancy d=200, n=10 | discrepancy: 0.273 -> 0.280 | Hurt |
| use Orthogonal instead of MC-antithetic | permutation sampling · Discrepancy d=200, n=1000 | discrepancy: 0.027 -> 0.023 | Improved |
| use Sobol instead of MC-antithetic | permutation sampling · Discrepancy d=50, n=1000 | discrepancy: 0.027 -> 0.022 | Improved |
| use herding instead of Orthogonal/Sobol for large n and d | permutation sampling · Runtime d=200, n=1000 | runtime (s): 242.516 vs 1.996 (Sobol) | Hurt |
| use herding or SBQ instead of MC-antithetic | Shapley value estimation, GBDT and MLP · cal_housing and make_regression (low-dim) | MSE vs exact Shapley values | Improved |
| use herding or SBQ instead of MC-antithetic | Shapley value estimation, GBDT · adult and year (high-dim) | MSE vs exact Shapley values | Hurt |
| use SBQ instead of herding | Shapley value estimation, GBDT · low-dimensional tabular datasets | MSE vs exact Shapley values | Improved |
| use Orthogonal instead of MC-antithetic | Shapley value estimation, GBDT · six tabular datasets | MSE vs exact Shapley values | Improved |
| use Sobol instead of MC-antithetic | Shapley value estimation, GBDT · six tabular datasets | MSE vs exact Shapley values | Mixed |
| use Sobol instead of MC-antithetic | Shapley value estimation, ResNet50 on ImageNet · 256-tile image Shapley values | MSE estimate | Improved |
Closest papers by meaning
- Regression-adjusted Monte Carlo Estimators for Shapley Values and Probabilistic Values (NeurIPS 2025 · similarity 0.91)
- Provably Accurate Shapley Value Estimation via Leverage Score Sampling (ICLR 2025 · similarity 0.88)
- DU-Shapley: A Shapley Value Proxy for Efficient Dataset Valuation (NeurIPS 2024 · similarity 0.87)
- A Unified Framework for Provably Efficient Algorithms to Estimate Shapley Values (NeurIPS 2025 · similarity 0.87)
- shapiq: Shapley Interactions for Machine Learning (NeurIPS 2024 · similarity 0.86)
- Amortized Linear-time Exact Shapley Value for Product-Kernel Methods (NeurIPS 2026 · 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.
