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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

We propose interpretable graph neural networks for sampling and recovery ofgraph signals, respectively. To take informative measurements, we propose a newgraph neural sampling module, which aims to select those vertices thatmaximally express their corresponding neighborhoods. Such expressiveness can bequantified by the mutual information between vertices' features andneighborhoods' features, which are estimated via a graph neural network. Toreconstruct an original graph signal from the sampled measurements, we proposea graph neural recovery module based on the algorithm-unrolling technique.Compared to previous analytical sampling and recovery, the proposed methods areable to flexibly learn a variety of graph signal models from data by leveragingthe learning ability of neural networks; compared to previousneural-network-based sampling and recovery, the proposed methods are designedthrough exploiting specific graph properties and provide interpretability. Wefurther design a new multiscale graph neural network, which is a trainablemultiscale graph filter bank and can handle various graph-related learningtasks. The multiscale network leverages the proposed graph neural sampling andrecovery modules to achieve multiscale representations of a graph. In theexperiments, we illustrate the effects of the proposed graph neural samplingand recovery modules and find that the modules can flexibly adapt to variousgraph structures and graph signals. In the task of active-sampling-basedsemi-supervised learning, the graph neural sampling module improves theclassification accuracy over 10% in Cora dataset. We further validate theproposed multiscale graph neural network on several standard datasets for bothvertex and graph classification. The results show that our method consistentlyimproves the classification accuracies.

Sampling and Recovery of Graph Signals based on Graph Neural Networks