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

Node classification is the task of predicting the labels of unlabeled nodesin a graph. State-of-the-art methods based on graph neural networks achieveexcellent performance when all labels are available during training. But inreal-life, models are often applied on data with new classes, which can lead tomassive misclassification and thus significantly degrade performance. Hence,developing open-set classification methods is crucial to determine if a givensample belongs to a known class. Existing methods for open-set nodeclassification generally use transductive learning with part or all of thefeatures of real unseen class nodes to help with open-set classification. Inthis paper, we propose a novel generative open-set node classification method,i.e. $\mathcal{G}^2Pxy$, which follows a stricter inductive learning settingwhere no information about unknown classes is available during training andvalidation. Two kinds of proxy unknown nodes, inter-class unknown proxies andexternal unknown proxies are generated via mixup to efficiently anticipate thedistribution of novel classes. Using the generated proxies, a closed-setclassifier can be transformed into an open-set one, by augmenting it with anextra proxy classifier. Under the constraints of both cross entropy loss andcomplement entropy loss, $\mathcal{G}^2Pxy$ achieves superior effectiveness forunknown class detection and known class classification, which is validated byexperiments on benchmark graph datasets. Moreover, $\mathcal{G}^2Pxy$ does nothave specific requirement on the GNN architecture and shows goodgeneralizations.

$\mathcal{G}^2Pxy$: Generative Open-Set Node Classification on Graphs with Proxy Unknowns