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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Most of the existing Out-Of-Distribution (OOD) detection algorithms depend onsingle input source: the feature, the logit, or the softmax probability.However, the immense diversity of the OOD examples makes such methods fragile.There are OOD samples that are easy to identify in the feature space while hardto distinguish in the logit space and vice versa. Motivated by thisobservation, we propose a novel OOD scoring method named Virtual-logit Matching(ViM), which combines the class-agnostic score from feature space and theIn-Distribution (ID) class-dependent logits. Specifically, an additional logitrepresenting the virtual OOD class is generated from the residual of thefeature against the principal space, and then matched with the original logitsby a constant scaling. The probability of this virtual logit after softmax isthe indicator of OOD-ness. To facilitate the evaluation of large-scale OODdetection in academia, we create a new OOD dataset for ImageNet-1K, which ishuman-annotated and is 8.8x the size of existing datasets. We conductedextensive experiments, including CNNs and vision transformers, to demonstratethe effectiveness of the proposed ViM score. In particular, using the BiT-Smodel, our method gets an average AUROC 90.91% on four difficult OODbenchmarks, which is 4% ahead of the best baseline. Code and dataset areavailable at https://github.com/haoqiwang/vim.

ViM: Out-Of-Distribution with Virtual-logit Matching