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

In recent years, deep models have achieved remarkable success in variousvision tasks. However, their performance heavily relies on large trainingdatasets. In contrast, humans exhibit hybrid learning, seamlessly integratingstructured knowledge for cross-domain recognition or relying on a smalleramount of data samples for few-shot learning. Motivated by this human-likeepistemic process, we aim to extend hybrid learning to computer vision tasks byintegrating structured knowledge with data samples for more effectiverepresentation learning. Nevertheless, this extension faces significantchallenges due to the substantial gap between structured knowledge and deepfeatures learned from data samples, encompassing both dimensions and knowledgegranularity. In this paper, a novel Epistemic Graph Layer (EGLayer) isintroduced to enable hybrid learning, enhancing the exchange of informationbetween deep features and a structured knowledge graph. Our EGLayer is composedof three major parts, including a local graph module, a query aggregationmodel, and a novel correlation alignment loss function to emulate humanepistemic ability. Serving as a plug-and-play module that can replace thestandard linear classifier, EGLayer significantly improves the performance ofdeep models. Extensive experiments demonstrates that EGLayer can greatlyenhance representation learning for the tasks of cross-domain recognition andfew-shot learning, and the visualization of knowledge graphs can aid in modelinterpretation.

Epistemic Graph: A Plug-And-Play Module For Hybrid Representation Learning