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

Self-supervised Learning (SSL) aims at learning representations of objectswithout relying on manual labeling. Recently, a number of SSL methods for graphrepresentation learning have achieved performance comparable to SOTAsemi-supervised GNNs. A Siamese network, which relies on data augmentation, isthe popular architecture used in these methods. However, these methods rely onheuristically crafted data augmentation techniques. Furthermore, they useeither contrastive terms or other tricks (e.g., asymmetry) to avoid trivialsolutions that can occur in Siamese networks. In this study, we propose,GraphSurgeon, a novel SSL method for GNNs with the following features. First,instead of heuristics we propose a learnable data augmentation method that isjointly learned with the embeddings by leveraging the inherent signal encodedin the graph. In addition, we take advantage of the flexibility of thelearnable data augmentation and introduce a new strategy that augments in theembedding space, called post augmentation. This strategy has a significantlylower memory overhead and run-time cost. Second, as it is difficult to sampletruly contrastive terms, we avoid explicit negative sampling. Third, instead ofrelying on engineering tricks, we use a scalable constrained optimizationobjective motivated by Laplacian Eigenmaps to avoid trivial solutions. Tovalidate the practical use of GraphSurgeon, we perform empirical evaluationusing 14 public datasets across a number of domains and ranging from small tolarge scale graphs with hundreds of millions of edges. Our finding shows thatGraphSurgeon is comparable to six SOTA semi-supervised and on par with fiveSOTA self-supervised baselines in node classification tasks. The source code isavailable at https://github.com/zekarias-tilahun/graph-surgeon.

Jointly Learnable Data Augmentations for Self-Supervised GNNs