AI Paper Atlas

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

Improved Gradient-Based Adversarial Attacks for Quantized Networks

Kartik Gupta, Thalaiyasingam Ajanthan

AAAI 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Provable robust learning guarantees · 16 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 temperature scaling with NJS to PGDBNN-WQ ResNet-18 · CIFAR-10adversarial accuracy: 26.98 -> 0.00Improved
Add temperature scaling with HNS to PGDBNN-WQ ResNet-18 · CIFAR-10adversarial accuracy: 26.98 -> 0.00Improved
Add temperature scaling with NJS to PGDBNN-WAQ ResNet-18 · CIFAR-10adversarial accuracy: 8.57 -> 0.03Improved
Add temperature scaling with HNS to PGDBNN-WAQ ResNet-18 · CIFAR-10adversarial accuracy: 8.57 -> 0.04Improved
Add temperature scaling with NJS to PGDBNN-WQ ResNet-18 · CIFAR-100adversarial accuracy: 8.23 -> 2.45Improved
Add temperature scaling with HNS to PGDBNN-WQ ResNet-18 · CIFAR-100adversarial accuracy: 8.23 -> 0.00Improved
Add temperature scaling to PGDAdversarially trained floating-point ResNet-18 · CIFAR-10adversarial accuracy: 48.73 -> 47.17 (NJS); 48.73 -> 48.54 (HNS)Mixed
Add temperature scaling to PGDAdversarially trained BNN-WQ ResNet-18 · CIFAR-10adversarial accuracy: 41.34 -> 40.76 (NJS); 41.34 -> 40.67 (HNS)Improved
Replace per-model temperature selection with fixed beta=0.1 (ablation)PGD on adversarially trained floating-point ResNet-18 · CIFAR-10adversarial accuracy: 48.73 -> 61.27Hurt
Replace per-model temperature selection with fixed beta=0.1 (ablation)FGSM on adversarially trained floating-point ResNet-18 · CIFAR-10adversarial accuracy: 62.38 -> 69.52Hurt
Use HNS rather than NJS for the one-step attack (ablation)BNN-WQ ResNet-18 · CIFAR-10adversarial accuracy: FGSM++ NJS 3.46; FGSM++ HNS 2.51Improved
Increase PGD iterations, attack radius, or step size (ablation)Original gradient-based attacks on BNNs · CIFAR-10adversarial accuracyNo 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