Paul-Gauthier Noé, Miquel Perelló-Nieto, Jean-François Bonastre, Peter Flach
{"title":"用夏普利合成解释单纯形上的概率预测","authors":"Paul-Gauthier Noé, Miquel Perelló-Nieto, Jean-François Bonastre, Peter Flach","doi":"arxiv-2408.01382","DOIUrl":null,"url":null,"abstract":"Originating in game theory, Shapley values are widely used for explaining a\nmachine learning model's prediction by quantifying the contribution of each\nfeature's value to the prediction. This requires a scalar prediction as in\nbinary classification, whereas a multiclass probabilistic prediction is a\ndiscrete probability distribution, living on a multidimensional simplex. In\nsuch a multiclass setting the Shapley values are typically computed separately\non each class in a one-vs-rest manner, ignoring the compositional nature of the\noutput distribution. In this paper, we introduce Shapley compositions as a\nwell-founded way to properly explain a multiclass probabilistic prediction,\nusing the Aitchison geometry from compositional data analysis. We prove that\nthe Shapley composition is the unique quantity satisfying linearity, symmetry\nand efficiency on the Aitchison simplex, extending the corresponding axiomatic\nproperties of the standard Shapley value. We demonstrate this proper multiclass\ntreatment in a range of scenarios.","PeriodicalId":501316,"journal":{"name":"arXiv - CS - Computer Science and Game Theory","volume":"41 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Explaining a probabilistic prediction on the simplex with Shapley compositions\",\"authors\":\"Paul-Gauthier Noé, Miquel Perelló-Nieto, Jean-François Bonastre, Peter Flach\",\"doi\":\"arxiv-2408.01382\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Originating in game theory, Shapley values are widely used for explaining a\\nmachine learning model's prediction by quantifying the contribution of each\\nfeature's value to the prediction. This requires a scalar prediction as in\\nbinary classification, whereas a multiclass probabilistic prediction is a\\ndiscrete probability distribution, living on a multidimensional simplex. In\\nsuch a multiclass setting the Shapley values are typically computed separately\\non each class in a one-vs-rest manner, ignoring the compositional nature of the\\noutput distribution. In this paper, we introduce Shapley compositions as a\\nwell-founded way to properly explain a multiclass probabilistic prediction,\\nusing the Aitchison geometry from compositional data analysis. We prove that\\nthe Shapley composition is the unique quantity satisfying linearity, symmetry\\nand efficiency on the Aitchison simplex, extending the corresponding axiomatic\\nproperties of the standard Shapley value. We demonstrate this proper multiclass\\ntreatment in a range of scenarios.\",\"PeriodicalId\":501316,\"journal\":{\"name\":\"arXiv - CS - Computer Science and Game Theory\",\"volume\":\"41 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-02\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Computer Science and Game Theory\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2408.01382\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Computer Science and Game Theory","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2408.01382","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Explaining a probabilistic prediction on the simplex with Shapley compositions
Originating in game theory, Shapley values are widely used for explaining a
machine learning model's prediction by quantifying the contribution of each
feature's value to the prediction. This requires a scalar prediction as in
binary classification, whereas a multiclass probabilistic prediction is a
discrete probability distribution, living on a multidimensional simplex. In
such a multiclass setting the Shapley values are typically computed separately
on each class in a one-vs-rest manner, ignoring the compositional nature of the
output distribution. In this paper, we introduce Shapley compositions as a
well-founded way to properly explain a multiclass probabilistic prediction,
using the Aitchison geometry from compositional data analysis. We prove that
the Shapley composition is the unique quantity satisfying linearity, symmetry
and efficiency on the Aitchison simplex, extending the corresponding axiomatic
properties of the standard Shapley value. We demonstrate this proper multiclass
treatment in a range of scenarios.