AI Paper Atlas

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

Improving DNN Robustness to Adversarial Attacks using Jacobian Regularization

Daniel Jakubovitz, Raja Giryes

ECCV 2018 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 205 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 Jacobian regularization as post-processingCNN · MNISTDeepFool rho_adv: 20.67 -> 34.24Improved
add Jacobian regularization as post-processingCNN · CIFAR-10DeepFool rho_adv: 1.21 -> 3.42Improved
add Jacobian regularization as post-processingCNN · CIFAR-100DeepFool rho_adv: 0.61 -> 1.55Improved
add adversarial training on top of Jacobian regularizationJacobian regularization · CIFAR-10DeepFool rho_adv: 3.42 -> 6.03Improved
apply Jacobian regularization as post-processing instead of during regular training (ablation)Jacobian regularization · MNISTDeepFool rho_adv: 32.89 -> 34.24Improved
regularize last-layer Jacobian instead of L-1 layer (fixed W) (ablation)Jacobian regularization · MNISTDeepFool rho_adv: 29.61 -> 34.24Improved
regularize last-layer Jacobian instead of L-2 layer (fixed weights) (ablation)Jacobian regularization · MNISTDeepFool rho_adv: 22.52 -> 34.24Improved
raise lambda beyond 0.1 (ablation)Jacobian regularization · MNISTDeepFool rho_advHurt
increase fraction of perturbed examples in adversarial training from 10% to 50% (ablation)Adversarial Training · MNISTDeepFool rho_adv: 23.73 -> 18.74Hurt
Jacobian regularization under JSMA attack versus Input Gradient regularizationJacobian regularization · MNISTJSMA test accuracyHurt
Jacobian regularization under JSMA attack versus other defensesJacobian regularization · CIFAR-10JSMA test accuracyImproved
add Jacobian regularizationCNN · MNISTJacobian Frobenius norm on perturbed inputs: 0.1877 -> 0.055Improved

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