AI Paper Atlas

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

Defensive Quantization: When Efficiency Meets Robustness

Ji Lin, Chuang Gan, Song Han

ICLR 2019 · Adversarial robustness and security › Certified and empirical robustness · approach: Regularized defenses against attacks · 33 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 Lipschitz regularization to 5-bit activation quantizationWide ResNet-28-10 · CIFAR-10 test, FGSM ε=8accuracy: 19.0 -> 51.8Improved
Add DQ to normal trainingWide ResNet-16-4 · SVHN, PGD ε=8accuracy: 05 -> 10Improved
Add DQ to normal trainingWide ResNet-28-10 · CIFAR-10, PGD ε=8accuracy: 01 -> 04Improved
Add DQ to R+FGSM adversarial trainingWide ResNet-16-4 · SVHN, PGD ε=8accuracy: 30 -> 47Improved
Add DQ to R+FGSM adversarial trainingWide ResNet-28-10 · CIFAR-10, PGD ε=8accuracy: 43 -> 50Improved
Add DQ to PGD adversarial trainingWide ResNet-28-10 · CIFAR-10, PGD ε=8accuracy: 44 -> 50Improved
Add DQ regularization to 4-bit ReLU1 quantizationResNet · CIFAR-10, cleanaccuracy: 93.49% -> 95.14%Improved
Add DQ regularization to 4-bit ReLU6 quantizationResNet · CIFAR-10, cleanaccuracy: 94.41% -> 95.09%Improved
Use Tanh-based rather than full-precision activation quantizationPGD-trained ResNet-18 · CIFAR-10 PGD white-boxaccuracy: 50.51% -> 71.03%Improved
Use Tanh-based rather than full-precision activation quantizationPGD-trained ResNet-18 · CIFAR-10 PGD black-boxaccuracy: 65.58% -> 61.57%Hurt
Use ReLU6 quantization rather than full precisionPGD-trained ResNet-18 · CIFAR-10 PGD black-boxaccuracy: 65.58% -> 59.44%Hurt
Replace adversarial FGSM training with R+FGSM training (ablation)Wide ResNet · CIFAR-10, FGSM black-box ε=8accuracy: 51.65% -> 87.39%Improved

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