AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

FastJSMA: Accelerating Jacobian-based Saliency Map Attacks through Gradient Decoupling

Zhenghao Gao, Shengjie Xu, Zijing Li, Meixi Chen, Chaojian Yu, Yuanjie Shao, Changxin Gao

ICCV 2025 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training with feature consistency · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add class excitation gradient g+ to the suppression-only attack (ablation)FastJSMA · CIFAR-10ASR (%): 35.75 -> 43.98Improved
add class suppression gradient g- to the excitation-only attack (ablation)FastJSMA · CIFAR-10ASR (%): 30.66 -> 43.98Improved
use suppression gradient alone instead of excitation gradient alone (ablation)FastJSMA · CIFAR-10ASR (%): 30.66 -> 35.75Improved
use both gradients instead of a single one: effect on generation time (ablation)FastJSMA · CIFAR-10time (µs): 339 vs 432 and 443No clear effect
replace full-Jacobian JSMANT with FastJSMA gradient decouplingJSMANT · CIFAR-10time (µs): 11,769 -> 339Improved
replace full-Jacobian JSMANT with FastJSMA gradient decouplingJSMANT · CIFAR-100time (µs): 61,166 -> 751Improved
replace full-Jacobian JSMANT with FastJSMA gradient decouplingJSMANT · CIFAR-100ASR (%): 43.87 -> 45.30Improved
use FastJSMA instead of weighted WJSMAWJSMA · CIFAR-10ASR (%): 51.56 -> 43.98Hurt
increase max attempt (class probing) from 1 to 10 (ablation)FastJSMA · CIFAR-10ASR (%): 35.2% at attempt=1 to 44.7% at attempt=10Improved
increase max attempt from 1 to 10: effect on time (ablation)FastJSMA · CIFAR-10time (µs): 314 -> 417Hurt
increase number of perturbed pixels from 2 to 50 (ablation)FastJSMA · CIFAR-10ASR (%): 26.0% with 2 pixels to 69.5% with 50 pixelsImproved
increase number of perturbed pixels: effect on time (ablation)FastJSMA · CIFAR-10time (s): 0.0003-0.0004 secondsNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy