Analysis of regression in game theory approach

By Stan Lipovetsky, Michael Conklin. 2001

652 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Statistics and Probability

Citations in the corpus

A Unified Approach to Interpreting Model Predictions (2017, methodology): “Classic Shapley value estimation: There are three previous methods that use classic equations from cooperative game theory to compute explanations of model predictions: Shapley regression values [3], Shapley sampling values [8], and Quantitative Input Influence [2].”
An unexpected unity among methods for interpreting model predictions (2016, methodology): “As noted by Shrikumar et Shapley regression values are an approach to computing feature importance in the presence of multicollinearity [2].”
Explainable AI for Trees: From Local Explanations to Global Understanding (2019, methodology): “Perhaps surprisingly, despite the seeming variety of different local explanation methods, two back propagation-style deep learning methods [3, 65], local linear mimic models [56], and several game theoretic methods [39, 71, 15] were recently unified into a single class of additive feature attributio”
Algorithms to estimate Shapley value feature attributions (2022, neutral): “Lipovetsky & Conklin [57], Štrumbelj et al.”
Infusing domain knowledge in AI-based "black box" models for better explainability with application in bankruptcy prediction (2019, methodology): “…approach called ”SHAP” which unifies seven previous approaches: LIME [60], DeepLIFT [61], Tree Interpreter [18], QII [30], Shapley sampling values [68], Shapley regression values [53], and Layer-wise relevance propagation [21] to make the explanation of prediction for any machine learning model.”
Understanding Global Feature Contributions With Additive Importance Measures (2020, methodology): “The mean SHAP value of the loss [24] and Shapley Net Effects for linear models [22] are both similar to SAGE, but they involve more expensive calculations (explaining every individual prediction, or ﬁtting an exponential number of linear models).”
Explainable Deep Learning Models in Medical Image Analysis (2020, methodology): “It was originally proposed for the analysis of regression [20].”
Determining Predictor Importance in Hierarchical Linear Models Using Dominance Analysis (2013, background): “…Merenda, and Gold (1980) as the squared semipartial correlations for each predictor in each subset model averaged over all orders of the predictors, by Chevan and Sutherland (1991) as the ‘‘independent component’’ of each predictor, and by Lipovetsky and Conklin (2001) as the ‘‘Shapley value.’’”
Neural Architecture Search with GBDT (2020, background): “4 SHAP value [12] is a uniﬁed measure of different Shapley values [9, 23, 5] which reﬂects the importance of features on the result considering their cooperation and interaction by solving a combined cooperative game theory problem [21].”
Shapley values for feature selection: The good, the bad, and the axioms (2021, neutral): “It was in this literature that the Shapley value was first applied to linear regression models, with its own history of discourse (see [2]–[7] and the more critical [8], which traces development to [9]–[11], with reinventions by [12] and [13]).”
Estimators of Relative Importance in Linear Regression Based on Variance Decomposition (2007, methodology): “Lipovetsky and Conklin (2001) reinvented LMG from a game-theoretic perspective by applying the Shapley value (Shapley 1953).”
Variable Importance Assessment in Regression: Linear Regression versus Random Forest (2009, methodology): “Strobl et al. (2008) positioned themselves at the conditional extreme by considering it “bias” that variables with the same coefficients receive different importances due to inter-regressor correlations; thus, following their logic, one would also have to reject the idea of predicting Y based on X2 ”
Explaining by Removing: A Unified Framework for Model Explanation (2020, methodology): “Finally, Shapley Net Effects (Lipovetsky and Conklin, 2001), SPVIM (Williamson and Feng, 2020) and the original IME (Štrumbelj et al.”
Problems with Shapley-value-based explanations as feature importance measures (2020, methodology): “A variety of methods to assign feature influence using the Shapley value have recently been developed (Lipovetsky & Conklin, 2001; Štrumbelj & Kononenko, 2014; Lundberg et al., 2018; Datta et al., 2016b; Merrick & Taly, 2019; Frye et al., 2019; Aas et al., 2019).”
On Interpretability of Artificial Neural Networks: A Survey (2020, methodology): “Furthermore, Shapley value from cooperative game theory was used in [6], [27], [39], [113], [115].”