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

The decoupled Graph Convolutional Network (GCN), a recent development of GCNthat decouples the neighborhood aggregation and feature transformation in eachconvolutional layer, has shown promising performance for graph representationlearning. Existing decoupled GCNs first utilize a simple neural network (e.g.,MLP) to learn the hidden features of the nodes, then propagate the learnedfeatures on the graph with fixed steps to aggregate the information ofmulti-hop neighborhoods. Despite effectiveness, the aggregation operation,which requires the whole adjacency matrix as the input, is involved in themodel training, causing high training cost that hinders its potential on largergraphs. On the other hand, due to the independence of node attributes as theinput, the neural networks used in decoupled GCNs are very simple, and advancedtechniques cannot be applied to the modeling. To this end, we further liberatethe aggregation operation from the decoupled GCN and propose a new paradigm ofGCN, termed Neighborhood Convolutional Network (NCN), that utilizes theneighborhood aggregation result as the input, followed by a specialconvolutional neural network tailored for extracting expressive noderepresentations from the aggregation input. In this way, the model couldinherit the merit of decoupled GCN for aggregating neighborhood information, atthe same time, develop much more powerful feature learning modules. A trainingstrategy called mask training is incorporated to further boost the modelperformance. Extensive results demonstrate the effectiveness of our model forthe node classification task on diverse homophilic graphs and heterophilicgraphs.

Neighborhood Convolutional Network: A New Paradigm of Graph Neural Networks for Node Classification