KIT | KIT-Bibliothek | Impressum | Datenschutz

A Step Towards Global Counterfactual Explanations: Approximating the Feature Space Through Hierarchical Division and Graph Search

Becker, Maximilian ORCID iD icon; Burkart, Nadia ORCID iD icon; Birnstill, Pascal; Beyerer, Jürgen

Abstract (englisch):

The field of Explainable Artificial Intelligence (XAI) tries to make learned models more understandable. One type of explanation for such models are counterfactual explanations. Counterfactual explanations explain the decision for a specific instance, the factual, by providing a similar instance which leads to a different decision, the counterfactual. In this work a new approaches around the idea of counterfactuals was developed. It generates a data structure over the feature space of a classification problem to accelerate the search for counterfactuals and augments them with global explanations. The approach maps the feature space by hierarchically dividing it into regions which belong to the same class. It is applicable in any case where predictions can be generated for input data, even without direct access to the model. The framework works well for lower-dimensional problems but becomes unpractical due to high computation times at around 12 to 15 dimensions.

Verlagsausgabe §
DOI: 10.5445/IR/1000139219/pub
Veröffentlicht am 27.01.2022
Postprint §
DOI: 10.5445/IR/1000139219
Veröffentlicht am 12.08.2022
DOI: 10.54364/aaiml.2021.1107
Zitationen: 3
Zitationen: 1
Cover der Publikation
Zugehörige Institution(en) am KIT Fraunhofer-Institut für Optronik, Systemtechnik und Bildauswertung (IOSB)
Institut für Anthropomatik und Robotik (IAR)
Institut für Informationssicherheit und Verlässlichkeit (KASTEL)
Publikationstyp Zeitschriftenaufsatz
Publikationsjahr 2021
Sprache Englisch
Identifikator ISSN: 2582-9793
KITopen-ID: 1000139219
HGF-Programm 46.23.04 (POF IV, LK 01) Engineering Security for Production Systems
Erschienen in Advances in Artificial Intelligence and Machine Learning
Band 1
Heft 2
Seiten 90-110
Vorab online veröffentlicht am 11.08.2021
Schlagwörter XAI, Counterfactual Explanations, Global Explanations
Nachgewiesen in Dimensions
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page