AI Paper Atlas

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

Universal Adversarial Perturbation via Prior Driven Uncertainty Approximation

Hong Liu, Rongrong Ji, Jie Li, Baochang Zhang, Yue Gao, Yongjian Wu, Feiyue Huang

ICCV 2019 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training refinements · 98 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 virtual epistemic uncertainty (MC dropout activation loss) (ablation)GD-UAP · ImageNet val, VGG-Ffooling rate (%): 85.96 -> 87.75Improved
add virtual epistemic uncertainty (MC dropout activation loss) (ablation)GD-UAP · ImageNet val, GoogleNetfooling rate (%): 51.61 -> 61.41Improved
add texture-bias prior and style loss (ablation)GD-UAP · ImageNet val, ResNet-50fooling rate (%): 35.59 -> 63.71Improved
add texture-bias prior and style loss (ablation)GD-UAP · ImageNet val, VGG-16fooling rate (%): 45.47 -> 51.63Improved
add texture-bias prior on top of epistemic uncertainty (PD-UA vs UA) (ablation)UA · ImageNet val, ResNet-50fooling rate (%): 38.87 -> 65.84Improved
add epistemic uncertainty to GD-UAP+P to get PD-UA (ablation)GD-UAP+P · ImageNet val, GoogleNetfooling rate (%): 58.37 -> 67.12Improved
use a texture image directly as the perturbation (no training)PP · ImageNet val, ResNet-150fooling rate (%): 28.87 -> 24.12Mixed
use texture prior initialization instead of pre-trained-perturbation prior (ablation)PD-UA · ImageNet val, GoogleNet (Table 4)fooling rate (%): 60.58 -> 65.23Improved
use low-pass filter instead of high-pass filter in LPFM (ablation)PD-UA · ImageNet val, ResNet-50fooling rate (%): 49.53 -> 65.84Improved
use LPFM instead of l2-norm gradient normalization (ablation)PD-UA · ImageNet val, VGG-16fooling rate (%): 26.31 -> 53.09Improved
increase pyramid levels from 1 to 3 (ablation)LPFM · ImageNet val, VGG-16fooling rate (%): 50.15 -> 53.09Improved
increase MC dropout samples T beyond 10 (ablation)PD-UA · ImageNet val, ResNet-50fooling rate (%)No 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