Adversarial explanations for knowledge graph embeddings

Betz, Patrick ; Meilicke, Christian ; Stuckenschmidt, Heiner

Document Type: Conference or workshop publication
Year of publication: 2022
Book title: Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence : Vienna, 23-29 July 2022
Page range: 2820-2826
Conference title: Thirty-First International Joint Conference on Artificial Intelligence
Location of the conference venue: Wien, Austria
Date of the conference: 23.-29.07.2022
Publisher: De Raedt, Luc
Place of publication: Darmstadt ; Vienna
Publishing house: International Joint Conferences on Artificial Intelligence
ISBN: 978-1-956792-00-3
Related URLs:
Publication language: English
Institution: School of Business Informatics and Mathematics > Praktische Informatik II (Stuckenschmidt 2009-)
Subject: 004 Computer science, internet
Abstract: We propose a novel black-box approach for performing adversarial attacks against knowledge graph embedding models. An adversarial attack is a small perturbation of the data at training time to cause model failure at test time. We make use of an efficient rule learning approach and use abductive reasoning to identify triples which are logical explanations for a particular prediction. The proposed attack is then based on the simple idea to suppress or modify one of the triples in the most confident explanation. Although our attack scheme is model independent and only needs access to the training data, we report results on par with state-of-the-art white-box attack methods that additionally require full access to the model architecture, the learned embeddings, and the loss functions. This is a surprising result which indicates that knowledge graph embedding models can partly be explained post hoc with the help of symbolic methods.

Dieser Eintrag ist Teil der Universitätsbibliographie.

Metadata export


+ Search Authors in

+ Page Views

Hits per month over past year

Detailed information

You have found an error? Please let us know about your desired correction here: E-Mail

Actions (login required)

Show item Show item