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

Graph Neural Networks (GNNs) have achieved notable success in learning fromgraph-structured data, owing to their ability to capture intricate dependenciesand relationships between nodes. They excel in various applications, includingsemi-supervised node classification, link prediction, and graph generation.However, it is important to acknowledge that the majority of state-of-the-artGNN models are built upon the assumption of an in-distribution setting, whichhinders their performance on real-world graphs with dynamic structures. In thisarticle, we aim to assess the impact of training GNNs on localized subsets ofthe graph. Such restricted training data may lead to a model that performs wellin the specific region it was trained on but fails to generalize and makeaccurate predictions for the entire graph. In the context of graph-basedsemi-supervised learning (SSL), resource constraints often lead to scenarioswhere the dataset is large, but only a portion of it can be labeled, affectingthe model's performance. This limitation affects tasks like anomaly detectionor spam detection when labeling processes are biased or influenced by humansubjectivity. To tackle the challenges posed by localized training data, weapproach the problem as an out-of-distribution (OOD) data issue by by aligningthe distributions between the training data, which represents a small portionof labeled data, and the graph inference process that involves makingpredictions for the entire graph. We propose a regularization method tominimize distributional discrepancies between localized training data and graphinference, improving model performance on OOD data. Extensive tests on popularGNN models show significant performance improvement on three citation GNNbenchmark datasets. The regularization approach effectively enhances modeladaptation and generalization, overcoming challenges posed by OOD data.

Addressing the Impact of Localized Training Data in Graph Neural Networks