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

Adversarial examples, generated by carefully crafted perturbation, haveattracted considerable attention in research fields. Recent works have arguedthat the existence of the robust and non-robust features is a primary cause ofthe adversarial examples, and investigated their internal interactions in thefeature space. In this paper, we propose a way of explicitly distilling featurerepresentation into the robust and non-robust features, using InformationBottleneck. Specifically, we inject noise variation to each feature unit andevaluate the information flow in the feature representation to dichotomizefeature units either robust or non-robust, based on the noise variationmagnitude. Through comprehensive experiments, we demonstrate that the distilledfeatures are highly correlated with adversarial prediction, and they havehuman-perceptible semantic information by themselves. Furthermore, we presentan attack mechanism intensifying the gradient of non-robust features that isdirectly related to the model prediction, and validate its effectiveness ofbreaking model robustness.

Distilling Robust and Non-Robust Features in Adversarial Examples by Information Bottleneck