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

An important aspect of developing reliable deep learning systems is devisingstrategies that make these systems robust to adversarial attacks. There is along line of work that focuses on developing defenses against these attacks,but recently, researchers have began to study ways to reverse engineer theattack process. This allows us to not only defend against several attackmodels, but also classify the threat model. However, there is still a lack oftheoretical guarantees for the reverse engineering process. Current approachesthat give any guarantees are based on the assumption that the data lies in aunion of linear subspaces, which is not a valid assumption for more complexdatasets. In this paper, we build on prior work and propose a novel frameworkfor reverse engineering of deceptions which supposes that the clean data liesin the range of a GAN. To classify the signal and attack, we jointly solve aGAN inversion problem and a block-sparse recovery problem. For the first timein the literature, we provide deterministic linear convergence guarantees forthis problem. We also empirically demonstrate the merits of the proposedapproach on several nonlinear datasets as compared to state-of-the-art methods.

A Linearly Convergent GAN Inversion-based Algorithm for Reverse Engineering of Deceptions