English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Despite the increasing relevance of explainable AI, assessing the quality ofexplanations remains a challenging issue. Due to the high costs associated withhuman-subject experiments, various proxy metrics are often used toapproximately quantify explanation quality. Generally, one possibleinterpretation of the quality of an explanation is its inherent value forteaching a related concept to a student. In this work, we extend artificialsimulatability studies to the domain of graph neural networks. Instead ofcostly human trials, we use explanation-supervisable graph neural networks toperform simulatability studies to quantify the inherent usefulness ofattributional graph explanations. We perform an extensive ablation study toinvestigate the conditions under which the proposed analyses are mostmeaningful. We additionally validate our methods applicability on real-worldgraph classification and regression datasets. We find that relevantexplanations can significantly boost the sample efficiency of graph neuralnetworks and analyze the robustness towards noise and bias in the explanations.We believe that the notion of usefulness obtained from our proposedsimulatability analysis provides a dimension of explanation quality that islargely orthogonal to the common practice of faithfulness and has greatpotential to expand the toolbox of explanation quality assessments,specifically for graph explanations.

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