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

Out-of-distribution (OOD) detection plays a crucial role in ensuring thesecurity of neural networks. Existing works have leveraged the fact thatIn-distribution (ID) samples form a subspace in the feature space, achievingstate-of-the-art (SOTA) performance. However, the comprehensive characteristicsof the ID subspace still leave under-explored. Recently, the discovery ofNeural Collapse ($\mathcal{NC}$) sheds light on novel properties of the IDsubspace. Leveraging insight from $\mathcal{NC}$, we observe that the PrincipalAngle between the features and the ID feature subspace forms a superiorrepresentation for measuring the likelihood of OOD. Building upon thisobservation, we propose a novel $\mathcal{NC}$-inspired OOD scoring function,named Entropy-enhanced Principal Angle (EPA), which integrates both the globalcharacteristic of the ID subspace and its inner property. We experimentallycompare EPA with various SOTA approaches, validating its superior performanceand robustness across different network architectures and OOD datasets.

EPA: Neural Collapse Inspired Robust Out-of-Distribution Detector