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

Clustering techniques attempt to group objects with similar properties into acluster. Clustering the nodes of an attributed graph, in which each node isassociated with a set of feature attributes, has attracted significantattention. Graph convolutional networks (GCNs) represent an effective approachfor integrating the two complementary factors of node attributes and structuralinformation for attributed graph clustering. However, oversmoothing of GCNsproduces indistinguishable representations of nodes, such that the nodes in agraph tend to be grouped into fewer clusters, and poses a challenge due to theresulting performance drop. In this study, we propose a smoothness sensor forattributed graph clustering based on adaptive smoothness-transition graphconvolutions, which senses the smoothness of a graph and adaptively terminatesthe current convolution once the smoothness is saturated to preventoversmoothing. Furthermore, as an alternative to graph-level smoothness, anovel fine-gained node-wise level assessment of smoothness is proposed, inwhich smoothness is computed in accordance with the neighborhood conditions ofa given node at a certain order of graph convolution. In addition, aself-supervision criterion is designed considering both the tightness withinclusters and the separation between clusters to guide the whole neural networktraining process. Experiments show that the proposed methods significantlyoutperform 12 other state-of-the-art baselines in terms of three differentmetrics across four benchmark datasets. In addition, an extensive study revealsthe reasons for their effectiveness and efficiency.

Smoothness Sensor: Adaptive Smoothness-Transition Graph Convolutions for Attributed Graph Clustering