AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

Attacking Binarized Neural Networks

Angus Galloway, Graham W. Taylor, Medhat Moussa

ICLR 2018 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 22 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
Use binary rather than full-precision weights and activations without adversarial trainingMNIST CNN, 256 first-layer kernels · MNIST test, FGSM ε=0.3accuracy: 17±3% -> 78±1%Improved
Add a learned scalar after hidden-layer ReLUs to the binary model (ablation)MNIST CNN, 256 first-layer kernels · MNIST clean testerror rate: 2.3±0.2% -> 1.2±0.1%Improved
Train the scaled binary model with PGD rather than no adversarial training (ablation)MNIST CNN, 256 first-layer kernels · MNIST test, FGSM ε=0.3accuracy: 16±3% -> 85.6±0.6%Improved
Add PGD adversarial training to a full-precision model (ablation)MNIST CNN, 256 first-layer kernels · MNIST test, FGSM ε=0.3accuracy: 17±3% -> 85±1%Improved
Stochastically quantize weights instead of deterministic binary quantization (ablation)MNIST binary CNN · MNIST targeted CW L2, 100 iterationsaccuracy: 22±3% -> 71±2%Improved
Increase binary-model capacity from 32 to 256 first-layer kernels (ablation)MNIST binary CNN · MNIST targeted CW L2, 100 iterationsaccuracy: 7±1% -> 22±3%Improved
Add FGSM adversarial training to deterministic binary models (ablation)MNIST binary CNN · MNIST test, FGSM ε=0.3accuracy: 78±1% -> 58±2%Hurt
Add FGSM adversarial training to scaled binary models (ablation)CIFAR-10 CNN, 256 first-layer kernels · CIFAR-10 black-box transfer, FGSM ε=0.3transfer accuracy: 21±1% -> 27.6±0.9%Mixed
Use PGD training rather than FGSM training on the scaled binary model (ablation)MNIST binary CNN, 256 first-layer kernels · MNIST black-box transfer, FGSM ε=0.3accuracy: 70±2% -> 96.4±0.4%Improved
Continue the targeted CW L2 attack from 100 to 1000 iterationsDeterministic binary and FGSM-adversarially-trained binary models · MNIST targeted CW L2accuracy: B and B+ reach 19±3% at 1000 iterationsHurt
Use stochastic quantization rather than deterministic binarization under a 1000-step CW L2 attack (ablation)MNIST binary CNN · MNIST targeted CW L2, 1000 iterationsaccuracy: 19±3% -> 38±3%Improved
Use PGD-trained scaled binary rather than PGD-trained full-precision modelsMNIST CNN, 256 first-layer kernels · MNIST test, FGSM ε=0.2accuracy: 92.9±0.3% -> 93.4±0.3%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