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

Deep Neural Networks (DNNs) are known to be vulnerable to adversarialattacks, i.e., an imperceptible perturbation to the input can mislead DNNstrained on clean images into making erroneous predictions. To tackle this,adversarial training is currently the most effective defense method, byaugmenting the training set with adversarial samples generated on the fly.Interestingly, we discover for the first time that there exist subnetworks withinborn robustness, matching or surpassing the robust accuracy of theadversarially trained networks with comparable model sizes, within randomlyinitialized networks without any model training, indicating that adversarialtraining on model weights is not indispensable towards adversarial robustness.We name such subnetworks Robust Scratch Tickets (RSTs), which are also bynature efficient. Distinct from the popular lottery ticket hypothesis, neitherthe original dense networks nor the identified RSTs need to be trained. Tovalidate and understand this fascinating finding, we further conduct extensiveexperiments to study the existence and properties of RSTs under differentmodels, datasets, sparsity patterns, and attacks, drawing insights regardingthe relationship between DNNs' robustness and theirinitialization/overparameterization. Furthermore, we identify the pooradversarial transferability between RSTs of different sparsity ratios drawnfrom the same randomly initialized dense network, and propose a Random RSTSwitch (R2S) technique, which randomly switches between different RSTs, as anovel defense method built on top of RSTs. We believe our findings about RSTshave opened up a new perspective to study model robustness and extend thelottery ticket hypothesis.

Drawing Robust Scratch Tickets: Subnetworks with Inborn Robustness Are Found within Randomly Initialized Networks