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

Although deep neural networks have shown promising performances on varioustasks, even achieving human-level performance on some, they are shown to besusceptible to incorrect predictions even with imperceptibly smallperturbations to an input. There exists a large number of previous works whichproposed to defend against such adversarial attacks either by robust inferenceor detection of adversarial inputs. Yet, most of them cannot effectively defendagainst whitebox attacks where an adversary has a knowledge of the model anddefense. More importantly, they do not provide a convincing reason why thegenerated adversarial inputs successfully fool the target models. To addressthese shortcomings of the existing approaches, we hypothesize that theadversarial inputs are tied to latent features that are susceptible toadversarial perturbation, which we call vulnerable features. Then based on thisintuition, we propose a minimax game formulation to disentangle the latentfeatures of each instance into robust and vulnerable ones, using variationalautoencoders with two latent spaces. We thoroughly validate our model for bothblackbox and whitebox attacks on MNIST, Fashion MNIST5, and Cat & Dog datasets,whose results show that the adversarial inputs cannot bypass our detectorwithout changing its semantics, in which case the attack has failed.

Learning to Disentangle Robust and Vulnerable Features for Adversarial Detection