Learning Model-Agnostic Counterfactual Explanations for Tabular Data

Martin Pawelczyk, Klaus Broelemann, Gjergji Kasneci

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

91 Scopus citations

Abstract

Counterfactual explanations can be obtained by identifying the smallest change made to an input vector to influence a prediction in a positive way from a user's viewpoint; for example, from 'loan rejected' to 'awarded' or from 'high risk of cardiovascular disease' to 'low risk'. Previous approaches would not ensure that the produced counterfactuals be proximate (i.e., not local outliers) and connected to regions with substantial data density (i.e., close to correctly classified observations), two requirements known as counterfactual faithfulness. Our contribution is twofold. First, drawing ideas from the manifold learning literature, we develop a framework, called C-CHVAE, that generates faithful counterfactuals. Second, we suggest to complement the catalog of counterfactual quality measures using a criterion to quantify the degree of difficulty for a certain counterfactual suggestion. Our real world experiments suggest that faithful counterfactuals come at the cost of higher degrees of difficulty.

Original languageEnglish
Title of host publicationThe Web Conference 2020 - Proceedings of the World Wide Web Conference, WWW 2020
PublisherAssociation for Computing Machinery, Inc
Pages3126-3132
Number of pages7
ISBN (Electronic)9781450370233
DOIs
StatePublished - 20 Apr 2020
Externally publishedYes
Event29th International World Wide Web Conference, WWW 2020 - Taipei, Taiwan, Province of China
Duration: 20 Apr 202024 Apr 2020

Publication series

NameThe Web Conference 2020 - Proceedings of the World Wide Web Conference, WWW 2020

Conference

Conference29th International World Wide Web Conference, WWW 2020
Country/TerritoryTaiwan, Province of China
CityTaipei
Period20/04/2024/04/20

Keywords

  • Counterfactual explanations
  • Interpretability
  • Transparency

Fingerprint

Dive into the research topics of 'Learning Model-Agnostic Counterfactual Explanations for Tabular Data'. Together they form a unique fingerprint.

Cite this