KIT | KIT-Bibliothek | Impressum | Datenschutz

Quantifying the Intrinsic Usefulness of Attributional Explanations for Graph Neural Networks with Artificial Simulatability Studies

Teufel, Jonas ORCID iD icon 1; Torresi, Luca 1; Friederich, Pascal ORCID iD icon 1
1 Institut für Theoretische Informatik (ITI), Karlsruher Institut für Technologie (KIT)


Despite the increasing relevance of explainable AI, assessing the quality of explanations remains a challenging issue. Due to the high costs associated with human-subject experiments, various proxy metrics are often used to approximately quantify explanation quality. Generally, one possible interpretation of the quality of an explanation is its inherent value for teaching a related concept to a student. In this work, we extend artificial simulatability studies to the domain of graph neural networks. Instead of costly human trials, we use explanation-supervisable graph neural networks to perform simulatability studies to quantify the inherent \textit{usefulness} of attributional graph explanations. We perform an extensive ablation study to investigate the conditions under which the proposed analyses are most meaningful. We additionally validate our method's applicability on real-world graph classification and regression datasets. We find that relevant explanations can significantly boost the sample efficiency of graph neural networks and analyze the robustness towards noise and bias in the explanations. We believe that the notion of usefulness obtained from our proposed simulatability analysis provides a dimension of explanation quality that is largely orthogonal to the common practice of faithfulness and has great potential to expand the toolbox of explanation quality assessments, specifically for graph explanations.

Verlagsausgabe §
DOI: 10.5445/IR/1000163744
Veröffentlicht am 03.11.2023
DOI: 10.1007/978-3-031-44067-0_19
Zitationen: 1
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Theoretische Informatik (ITI)
Publikationstyp Proceedingsbeitrag
Publikationsjahr 2023
Sprache Englisch
Identifikator ISBN: 978-3-031-44066-3
ISSN: 1865-0929
KITopen-ID: 1000163744
HGF-Programm 43.31.01 (POF IV, LK 01) Multifunctionality Molecular Design & Material Architecture
Erschienen in Explainable Artificial Intelligence. Part 2. Ed.: L. Longo. Proceedings. Pt. 2
Veranstaltung 1st World Conference On eXplainable Artificial Intelligence (xAI 2023), Lissabon, Portugal, 26.07.2023 – 28.07.2023
Verlag Springer Nature Switzerland
Seiten 361–381
Serie Communications in Computer and Information Science ; 1902
Vorab online veröffentlicht am 21.10.2023
Nachgewiesen in Dimensions
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page