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

We present an approach to quantifying both aleatoric and epistemicuncertainty for deep neural networks in image classification, based ongenerative adversarial networks (GANs). While most works in the literature thatuse GANs to generate out-of-distribution (OoD) examples only focus on theevaluation of OoD detection, we present a GAN based approach to learn aclassifier that produces proper uncertainties for OoD examples as well as forfalse positives (FPs). Instead of shielding the entire in-distribution datawith GAN generated OoD examples which is state-of-the-art, we shield each classseparately with out-of-class examples generated by a conditional GAN andcomplement this with a one-vs-all image classifier. In our experiments, inparticular on CIFAR10, CIFAR100 and Tiny ImageNet, we improve over the OoDdetection and FP detection performance of state-of-the-art GAN-training basedclassifiers. Furthermore, we also find that the generated GAN examples do notsignificantly affect the calibration error of our classifier and result in asignificant gain in model accuracy.

UQGAN: A Unified Model for Uncertainty Quantification of Deep Classifiers trained via Conditional GANs