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

Decision-based methods have shown to be effective in black-box adversarialattacks, as they can obtain satisfactory performance and only require to accessthe final model prediction. Gradient estimation is a critical step in black-boxadversarial attacks, as it will directly affect the query efficiency. Recentworks have attempted to utilize gradient priors to facilitate score-basedmethods to obtain better results. However, these gradient priors still sufferfrom the edge gradient discrepancy issue and the successive iteration gradientdirection issue, thus are difficult to simply extend to decision-based methods.In this paper, we propose a novel Decision-based Black-box Attack frameworkwith Gradient Priors (DBA-GP), which seamlessly integrates the data-dependentgradient prior and time-dependent prior into the gradient estimation procedure.First, by leveraging the joint bilateral filter to deal with each randomperturbation, DBA-GP can guarantee that the generated perturbations in edgelocations are hardly smoothed, i.e., alleviating the edge gradient discrepancy,thus remaining the characteristics of the original image as much as possible.Second, by utilizing a new gradient updating strategy to automatically adjustthe successive iteration gradient direction, DBA-GP can accelerate theconvergence speed, thus improving the query efficiency. Extensive experimentshave demonstrated that the proposed method outperforms other strong baselinessignificantly.

Boosting Decision-Based Black-Box Adversarial Attack with Gradient Priors