AI Paper Atlas

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

Towards Efficient Adversarial Training on Vision Transformers

Boxi Wu, Jindong Gu, Zhifeng Li, Deng Cai, Xiaofei He, Wei Liu

ECCV 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training with feature consistency · 31 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
random patch dropping (40%) added to Fast AT (ablation)Fast AT ViT-Base · ImageNet eps=2/255PGD-100 robust accuracy: 50.06 -> 36.81Hurt
attention-guided dropping (AGAT) added to Fast AT (ablation)Fast AT ViT-Base · ImageNet eps=2/255PGD-100 robust accuracy: 50.06 -> 51.11Improved
attention-guided dropping (AGAT) added to Fast AT (ablation)Fast AT ViT-Base · ImageNettraining time: 86H -> 56HImproved
attention-guided dropping (AGAT) added to Fast AT (ablation)Fast AT CaiT-S36 · ImageNettraining time: 76H -> 53HImproved
attention-guided dropping (AGAT) added to Fast AT (ablation)Fast AT LeViT-384 · ImageNet eps=2/255PGD-100 robust accuracy: 46.13 -> 48.59Improved
attention-guided dropping (AGAT) added to Fast AT (ablation)Fast AT ViT-Small · ImageNet eps=2/255PGD-100 robust accuracy: 49.41 -> 48.85No clear effect
attention-guided instead of random dropping at 60% computation (ablation)Random dropping ViT-Tiny · ImageNet eps=2/255PGD robust accuracy: 21.01 -> 46.60Improved
attention-guided instead of random dropping at 40% computation (ablation)Random dropping ViT-Base · ImageNet eps=2/255PGD robust accuracy: 36.81 -> 50.02Improved
increase AGAT dropped computation from 40% to 60% (ablation)AGAT ViT-Base · ImageNet eps=4/255PGD robust accuracy: 34.78 -> 32.90Hurt
random patch dropping added to Fast AT (ablation)Fast AT ViT-Tiny · ImageNet eps=2/255PGD-100 robust accuracy: 48.28 -> 31.16Hurt
evaluate AGAT under APGD-CE instead of PGD (ablation)AGAT ViT-Base · ImageNet eps=2/255robust accuracy: 50.01 -> 51.23No clear effect
add attention dropout to Fast ATFast AT ViTs · ImageNettraining stabilityImproved

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