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/

Zendy Plus

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

Zendy Tools

Zendy Open

Subgraph representation learning has emerged as an important problem, but itis by default approached with specialized graph neural networks on a largeglobal graph. These models demand extensive memory and computational resourcesbut challenge modeling hierarchical structures of subgraphs. In this paper, wepropose Subgraph-To-Node (S2N) translation, a novel formulation for learningrepresentations of subgraphs. Specifically, given a set of subgraphs in theglobal graph, we construct a new graph by coarsely transforming subgraphs intonodes. Demonstrating both theoretical and empirical evidence, S2N not onlysignificantly reduces memory and computational costs compared tostate-of-the-art models but also outperforms them by capturing both local andglobal structures of the subgraph. By leveraging graph coarsening methods, ourmethod outperforms baselines even in a data-scarce setting with insufficientsubgraphs. Our experiments on eight benchmarks demonstrate that fined-tunedmodels with S2N translation can process 183 -- 711 times more subgraph samplesthan state-of-the-art models at a better or similar performance level.

Translating Subgraphs to Nodes Makes Simple GNNs Strong and Efficient for Subgraph Representation Learning