AI Paper Atlas

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

LAFEAT: Piercing Through Adversarial Defenses With Latent Features

Yunrui Yu, Xitong Gao, Cheng-Zhong Xu

CVPR 2021 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training with feature consistency · 38 citations (OpenAlex)

Check 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
Use LAFEAT MT(1k,10) instead of AutoAttack AA100Adversarial weight perturbation defense · CIFAR-10accuracy under attack (%): 60.04 -> 59.94Hurt
Use LAFEAT MT(1k,10) instead of AutoAttack AA100Manifold regularization defense · CIFAR-10accuracy under attack (%): 1.35 -> 0.21Hurt
Use LAFEAT MT(1k,10) instead of AutoAttack AA100Adversarial weight perturbation defense · CIFAR-100accuracy under attack (%): 28.86 -> 28.77Hurt
Use LAFEAT MT(1k,10) instead of AutoAttack AA100Overfitting defense · CIFAR-100accuracy under attack (%): 18.95 -> 18.86Hurt
Add adversarial training of latent logits layers (ablation)PGD-7-trained WideResNet-32-10 · CIFAR-10clean accuracy (%): 79.56 -> 83.53Improved
Add adversarial training of latent logits layers (ablation)PGD-7-trained WideResNet-32-10 · CIFAR-10, PGD-100accuracy under attack (%): 47.06 -> 47.17Improved
Add adversarial training of latent logits layers (ablation)PGD-7-trained WideResNet-32-10 · CIFAR-10, LAFEAT --LF attackaccuracy under attack (%): 41.21 -> 41.50Improved
Add adversarial training of latent logits layers (ablation)PGD-7-trained WideResNet-32-10 · CIFAR-10, LAFEAT +LF attackaccuracy under attack (%): 41.05 -> 41.26Improved
Add latent features to an existing attack-tactic combination (ablation)LAFEAT · 14 defending modelsaccuracy under attack (%)Improved
Vary the interpolation weight between latent and output logits (ablation)LAFEAT · CIFAR-10 defense modelsaccuracy under attack (%)Mixed
Train auxiliary logits using the attack dataset rather than the training dataset (ablation)LAFEAT · Adversarial attack evaluationattack strengthNo 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