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), which learn the representation of a node byaggregating its neighbors, have become an effective computational tool indownstream applications. Over-smoothing is one of the key issues which limitthe performance of GNNs as the number of layers increases. It is because thestacked aggregators would make node representations converge toindistinguishable vectors. Several attempts have been made to tackle the issueby bringing linked node pairs close and unlinked pairs distinct. However, theyoften ignore the intrinsic community structures and would result in sub-optimalperformance. The representations of nodes within the same community/class needbe similar to facilitate the classification, while different classes areexpected to be separated in embedding space. To bridge the gap, we introducetwo over-smoothing metrics and a novel technique, i.e., differentiable groupnormalization (DGN). It normalizes nodes within the same group independently toincrease their smoothness, and separates node distributions among differentgroups to significantly alleviate the over-smoothing issue. Experiments onreal-world datasets demonstrate that DGN makes GNN models more robust toover-smoothing and achieves better performance with deeper GNNs.

Towards Deeper Graph Neural Networks with Differentiable Group Normalization