AI Paper Atlas

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

Adversarial Robustness vs. Model Compression, or Both?

Shaokai Ye, Kaidi Xu, Sijia Liu, Hao Cheng, Jan-Henrik Lambrechts, Huan Zhang, Aojun Zhou, Kaisheng Ma, Yanzhi Wang, Xue Lin

ICCV 2019 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training refinements · 149 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
prune a w=4 model down to w=1 with concurrent adversarial training instead of training w=1 adversarially from scratchLeNet · MNISTnatural / adversarial test accuracy: 11.35/11.35 -> 96.22/89.41Improved
start from w=8 instead of w=4 before pruning to w=1LeNet concurrent pruning · MNISTnatural / adversarial test accuracy: 96.22/89.41 -> 97.31/92.16Improved
start from very large w=16 rather than w=8 when pruning to w=1LeNet concurrent pruning · MNISTadversarial test accuracy: 92.16 -> 87.79Hurt
inherit initialization from larger adversarially trained model (lottery ticket) without pruning, w=4 to 2 (ablation)LeNet · MNISTnatural / adversarial test accuracy: 97.68/91.77 -> 11.35/11.35Hurt
inherited initialization from w=8 for w=2 model, no pruning (ablation)LeNet · MNISTnatural / adversarial test accuracy: 97.36/90.19Mixed
adversarial training of a small w=1 model instead of natural trainingLeNet · MNISTnatural test accuracy: 98.25 -> 11.35Hurt
adversarial training on CIFAR10 LeNet w=1 instead of natural trainingLeNet · CIFAR10natural / adversarial test accuracy: 74.84/0.01 -> 10.00/10.00Hurt
prune ResNet w=16 to w=4 versus training w=4 adversarially from scratchResNet · CIFAR10natural / adversarial test accuracy: 77.35/44.99 -> 78.85/47.19Improved
irregular pruning instead of filter pruning (ablation)VGG-16 and ResNet-18 · CIFAR10natural and adversarial accuracyImproved
column pruning instead of filter pruning (ablation)VGG-16 and ResNet-18 · CIFAR10natural and adversarial accuracyImproved
choose different initializations (kaiming normal, uniform) with Adam for adversarial training from scratch at w=1 (ablation)LeNet · MNISTtraining success: 1 out of 10 trials succeedsNo clear effect
apply concurrent pruned models against transfer attack from larger baseline modelspruned LeNet · MNISTadversarial test accuracy: 89.41 -> 93.55Improved

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