AI Paper Atlas

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

Integer-Arithmetic-Only Certified Robustness for Quantized Neural Networks

Haowen Lin, Jian Lou, Li Xiong, Cyrus Shahabi

ICCV 2021 · Adversarial robustness and security › Certified and empirical robustness · approach: Certified training and ensembles · 14 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
quantize to int8 with integer randomized smoothing instead of float networkrandomized smoothing, ResNet-50 · Caltech-101 sigma=1.00, radius 2.0certified accuracy: 0.41 -> 0.52Improved
use IntRS-quant instead of FloatRS-quant (rounded float noise on quantized net)quantized ResNet-20 · CIFAR-10 sigma=1.00, radius 0.25certified accuracy: 0.19 -> 0.35Improved
use IntRS-quant instead of FloatRS-quantquantized ResNet-50 · Caltech-101 sigma=1.00, radius 0.25certified accuracy: 0.20 -> 0.56Improved
quantize with IntRS versus full-precision float smoothing at low sigmaResNet-20 · CIFAR-10 sigma=0.50, radius 0.25certified accuracy: 0.54 -> 0.52Hurt
quantize with IntRS versus full-precision float smoothing at sigma=1.00ResNet-20 · CIFAR-10 sigma=1.00, radius 0.25certified accuracy: 0.39 -> 0.35Hurt
int8 integer-only inference instead of float inferenceIntRS · CIFAR-10 certification timeseconds per image: 194.4 -> 55.8Improved
int8 integer-only inference instead of float inferenceIntRS · Caltech-101 certification timeseconds per image: 159.9 -> 29.5Improved
increase noise sigma used in training and certification (ablation)IntRS-quant · CIFAR-10 sigma 0.12 -> 1.00certified accuracy at radius 0.25: 0.59 -> 0.35Hurt
certify with sigma=0.25 on model trained with sigma=0.12 (mismatch) (ablation)IntRS-quant · CIFAR-10certified accuracy at radius 0.25: 0.59 -> 0.36Hurt
IntRS under train/certify sigma mismatch versus Float-fpIntRS-quant · CIFAR-10 trained sigma=0.12, certified sigma=0.25certified accuracy at radius 0.25: 0.19 -> 0.36Improved
increase Monte-Carlo samples n from 100 to 10000 (ablation)IntRS-quant · CIFAR-10 sigma=0.25 predictionabstain rate: 0.15 -> 0.02Improved
increase Monte-Carlo samples n from 100 to 10000 (ablation)IntRS-quant · CIFAR-10 sigma=0.25 predictioncorrect rate: 0.73 -> 0.80Improved

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