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

Current out-of-distribution detection approaches usually present specialrequirements (e.g., collecting outlier data and hyperparameter validation) andproduce side effects (e.g., classification accuracy drop and slow/inefficientinferences). Recently, entropic out-of-distribution detection has been proposedas a seamless approach (i.e., a solution that avoids all previously mentioneddrawbacks). The entropic out-of-distribution detection solution uses the IsoMaxloss for training and the entropic score for out-of-distribution detection. TheIsoMax loss works as a drop-in replacement of the SoftMax loss (i.e., thecombination of the output linear layer, the SoftMax activation, and thecross-entropy loss) because swapping the SoftMax loss with the IsoMax lossrequires no changes in the model's architecture or trainingprocedures/hyperparameters. In this paper, we perform what we call anisometrization of the distances used in the IsoMax loss. Additionally, wepropose replacing the entropic score with the minimum distance score.Experiments showed that these modifications significantly increaseout-of-distribution detection performance while keeping the solution seamless.Besides being competitive with or outperforming all major current approaches,the proposed solution avoids all their current limitations, in addition tobeing much easier to use because only a simple loss replacement for trainingthe neural network is required. The code to replace the SoftMax loss with theIsoMax+ loss and reproduce the results is available athttps://github.com/dlmacedo/entropic-out-of-distribution-detection.

Enhanced Isotropy Maximization Loss: Seamless and High-Performance Out-of-Distribution Detection Simply Replacing the SoftMax Loss