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

Standard recognition approaches are unable to deal with novel categories attest time. Their overconfidence on the known classes makes the predictionsunreliable for safety-critical applications such as healthcare or autonomousdriving. Out-Of-Distribution (OOD) detection methods provide a solution byidentifying semantic novelty. Most of these methods leverage a learning stageon the known data, which means training (or fine-tuning) a model to capture theconcept of normality. This process is clearly sensitive to the amount ofavailable samples and might be computationally expensive for on-board systems.A viable alternative is that of evaluating similarities in the embedding spaceproduced by large pre-trained models without any further learning effort. Wefocus exactly on such a fine-tuning-free OOD detection setting. This workspresents an in-depth analysis of the recently introduced relational reasoningpre-training and investigates the properties of the learned embedding,highlighting the existence of a correlation between the inter-class featuredistance and the OOD detection accuracy. As the class separation depends on thechosen pre-training objective, we propose an alternative loss function tocontrol the inter-class margin, and we show its advantage with thoroughexperiments.

Large Class Separation is not what you need for Relational Reasoning-based OOD Detection