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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Uncovering rationales behind predictions of graph neural networks (GNNs) hasreceived increasing attention over recent years. Instance-level GNN explanationaims to discover critical input elements, like nodes or edges, that the targetGNN relies upon for making predictions. %These identified sub-structures canprovide interpretations of GNN's behavior. Though various algorithms areproposed, most of them formalize this task by searching the minimal subgraphwhich can preserve original predictions. However, an inductive bias isdeep-rooted in this framework: several subgraphs can result in the same orsimilar outputs as the original graphs. Consequently, they have the danger ofproviding spurious explanations and failing to provide consistent explanations.Applying them to explain weakly-performed GNNs would further amplify theseissues. To address this problem, we theoretically examine the predictions ofGNNs from the causality perspective. Two typical reasons for spuriousexplanations are identified: confounding effect of latent variables likedistribution shift, and causal factors distinct from the original input.Observing that both confounding effects and diverse causal rationales areencoded in internal representations, \tianxiang{we propose a new explanationframework with an auxiliary alignment loss, which is theoretically proven to beoptimizing a more faithful explanation objective intrinsically. Concretely forthis alignment loss, a set of different perspectives are explored: anchor-basedalignment, distributional alignment based on Gaussian mixture models,mutual-information-based alignment, etc. A comprehensive study is conductedboth on the effectiveness of this new framework in terms of explanationfaithfulness/consistency and on the advantages of these variants.

Faithful and Consistent Graph Neural Network Explanations with Rationale Alignment