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

Correctly classifying adversarial examples is an essential but challengingrequirement for safely deploying machine learning models. As reported inRobustBench, even the state-of-the-art adversarially trained models struggle toexceed 67% robust test accuracy on CIFAR-10, which is far from practical. Acomplementary way towards robustness is to introduce a rejection option,allowing the model to not return predictions on uncertain inputs, whereconfidence is a commonly used certainty proxy. Along with this routine, we findthat confidence and a rectified confidence (R-Con) can form two coupledrejection metrics, which could provably distinguish wrongly classified inputsfrom correctly classified ones. This intriguing property sheds light on usingcoupling strategies to better detect and reject adversarial examples. Weevaluate our rectified rejection (RR) module on CIFAR-10, CIFAR-10-C, andCIFAR-100 under several attacks including adaptive ones, and demonstrate thatthe RR module is compatible with different adversarial training frameworks onimproving robustness, with little extra computation. The code is available athttps://github.com/P2333/Rectified-Rejection.

Two Coupled Rejection Metrics Can Tell Adversarial Examples Apart