Higher order
- [1705.04977] Detecting Statistical Interactions from Neural Network Weights
- How does This Interaction Affect Me? Interpretable Attribution for Feature Interactions
- The Shapley Taylor Interaction Index
- Purifying Interaction Effects with the Functional ANOVA: An Efficient Algorithm for Recovering Identifiable Additive Models
- Neural Interaction Transparency (NIT): Disentangling Learned Interactions for Improved Interpretability
- [1806.05337] Hierarchical interpretations for neural network predictions
- Learning Global Additive Explanations for Neural Nets Using Model Distillation | OpenReview
- Higher-Order Explanations of Graph Neural Networks via Relevant Walks | IEEE Journals & Magazine | IEEE Xplore
Sobol and Shapley
- Sobol’ Indices and Shapley Value
- [KDD question] Shap values vs Sobol indices? · Issue #1382 · shap/shap · GitHub
- [1707.01334] Shapley effects for sensitivity analysis with correlated inputs: comparisons with Sobol' indices, numerical estimation and applications
- https://epubs.siam.org/doi/abs/10.1137/18M1221801
- https://www.esaim-proc.org/articles/proc/abs/2019/01/proc196511/proc196511.html
- Shapley Values for Feature Selection: The Good, the Bad, and the Axioms | IEEE Journals & Magazine | IEEE Xplore
- https://epubs.siam.org/doi/abs/10.1137/15M1048070
- Understanding Global Feature Contributions With Additive Importance Measures
- The Many Shapley Values for Model Explanation
ANOVA and Sobol
- ICML 22 [2206.09861] Additive Gaussian Processes Revisited
- [A Tutorial] https://www.sciencedirect.com/science/article/pii/S0010465509003087
- https://www.sciencedirect.com/science/article/pii/S0951832006001499
- AISTAT 20: Neural Decomposition: Functional ANOVA with Variational Autoencoders
- AISTAT 20: https://www.semanticscholar.org/paper/Purifying-Interaction-Effects-with-the-Functional-Lengerich-Tan/b7596df89a53f67a7bedd09f6c97efe263daf96c
- https://dl.acm.org/doi/abs/10.1016/j.envsoft.2022.105310
- https://www.semanticscholar.org/paper/Disentangling-by-Factorising-Kim-Mnih/04541599accc47d8174f63345ce9c987ef21685b
- https://www.semanticscholar.org/paper/The-Concrete-Distribution%3A-A-Continuous-Relaxation-Maddison-Mnih/515a21e90117941150923e559729c59f5fdade1c
- https://www.semanticscholar.org/paper/Bayesian-functional-ANOVA-modeling-using-Gaussian-Kaufman-Sain/05186498a81c026e6112f3b38cebcac2b1a6a211
- https://www.semanticscholar.org/paper/Functional-Data-Analysis-Müller/b29e5be7987216cb1aaf46c6927ff216e048240a
- https://www.semanticscholar.org/paper/Factorized-sparse-learning-models-with-high-order-Purushotham-Min/08864515e931bd69f5b321741eb5dcb7f90afbad
- https://www.semanticscholar.org/paper/pureGAM%3A-Learning-an-Inherently-Pure-Additive-Model-Sun-Wang/9aabc07f4f844b85b76c12270953733c818c1ad2
- https://www.semanticscholar.org/paper/Unifying-local-and-global-model-explanations-by-of-Hiabu-Meyer/632255597a6ff91e4c22a72c33f94a53687ea68a
- NeurIPS 21 Look at the Variance! Efficient Black-box Explanations with Sobol-based Sensitivity Analysis | OpenReview
- https://www.semanticscholar.org/paper/Decomposing-feature-level-variation-with-Covariate-Märtens-Campbell/0c67aeac626417845a442f94f48460b9c716d358
- https://www.semanticscholar.org/paper/A-Unified-Approach-to-Interpreting-Model-Lundberg-Lee/442e10a3c6640ded9408622005e3c2a8906ce4c2
- https://www.semanticscholar.org/paper/Learning-Global-Additive-Explanations-for-Neural-Tan-Caruana/02f90510adc84ec3fcc093e19b22e84809cdc2bc
- Uncovering pseudotemporal trajectories with covariates from single cell and bulk expression data