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 Convolutional Network (GCN) has been widely used in graph learningtasks. However, GCN-based models (GCNs) is an inherently coupled trainingframework repetitively conducting the complex neighboring aggregation, whichleads to the limitation of flexibility in processing large-scale graph. Withthe depth of layers increases, the computational and memory cost of GCNs growexplosively due to the recursive neighborhood expansion. To tackle theseissues, we present Node2Grids, a flexible uncoupled training framework thatleverages the independent mapped data for obtaining the embedding. Instead ofdirectly processing the coupled nodes as GCNs, Node2Grids supports a moreefficacious method in practice, mapping the coupled graph data into theindependent grid-like data which can be fed into the efficient ConvolutionalNeural Network (CNN). This simple but valid strategy significantly saves memoryand computational resource while achieving comparable results with the leadingGCN-based models. Specifically, by ranking each node's influence throughdegree, Node2Grids selects the most influential first-order as well assecond-order neighbors with central node fusion information to construct thegrid-like data. For further improving the efficiency of downstream tasks, asimple CNN-based neural network is employed to capture the significantinformation from the mapped grid-like data. Moreover, the grid-level attentionmechanism is implemented, which enables implicitly specifying the differentweights for neighboring nodes with different influences. In addition to thetypical transductive and inductive learning tasks, we also verify our frameworkon million-scale graphs to demonstrate the superiority of the proposedNode2Grids model against the state-of-the-art GCN-based approaches.

An Uncoupled Training Architecture for Large Graph Learning