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

Graphs neural networks (GNNs) have emerged as a powerful graph learning modeldue to their superior capacity in capturing critical graph patterns. To gaininsights about the model mechanism for interpretable graph learning, previousefforts focus on post-hoc local interpretation by extracting the data patternthat a pre-trained GNN model uses to make an individual prediction. However,recent works show that post-hoc methods are highly sensitive to modelinitialization and local interpretation can only explain the model predictionspecific to a particular instance. In this work, we address these limitationsby answering an important question that is not yet studied: how to provideglobal interpretation of the model training procedure? We formulate thisproblem as in-process global interpretation, which targets on distillinghigh-level and human-intelligible patterns that dominate the training procedureof GNNs. We further propose Graph Distribution Matching (GDM) to synthesizeinterpretive graphs by matching the distribution of the original andinterpretive graphs in the feature space of the GNN as its training proceeds.These few interpretive graphs demonstrate the most informative patterns themodel captures during training. Extensive experiments on graph classificationdatasets demonstrate multiple advantages of the proposed method, including highexplanation accuracy, time efficiency and the ability to reveal class-relevantstructure.

In-Process Global Interpretation for Graph Learning via Distribution Matching