AI Paper Atlas

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

Improving Adversarial Robustness via Guided Complement Entropy

Hao-Yun Chen, Jhao-Hong Liang, Shih-Chieh Chang, Jia-Yu Pan, Yu-Ting Chen, Wei Wei, Da-Cheng Juan

ICCV 2019 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 57 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
replace cross-entropy with GCE (alpha=1/3)ResNet-50 · Tiny ImageNeterror rate (%): 39.54 -> 38.56Improved
replace cross-entropy with GCE (alpha=1/3)LeNet-5 · MNISTerror rate (%): 0.8 -> 0.67Improved
replace cross-entropy with GCE (alpha=1/3)ResNet-56 · CIFAR10error rate (%): 7.99 -> 7.18Improved
replace cross-entropy with GCE (alpha=1/3)ResNet-56 · CIFAR100error rate (%): 31.9 -> 31.75No clear effect
use guided factor alpha=1/2 instead of smaller alpha (ablation)GCE · CIFAR100error rate (%): 31.75 -> 40.59Hurt
use guided factor alpha=1/2 instead of smaller alpha (ablation)GCE · CIFAR10error rate (%): 7.18 -> 9.18Hurt
train with GCE instead of XE, FGSM attackLeNet-5 · MNIST FGSM eps=0.2accuracy under attack (%): 38.88 -> 62.74Improved
train with GCE instead of XE, BIM attackLeNet-5 · MNIST BIM eps=0.2accuracy under attack (%): 2.15 -> 34.49Improved
train with GCE instead of XE, MIM attackLeNet-5 · MNIST MIM eps=0.2accuracy under attack (%): 2.29 -> 39.81Improved
train with GCE instead of XE, PGD attackResNet-56 · CIFAR10 PGD eps=0.04accuracy under attack (%): 0.0 -> 5.91Improved
train with GCE instead of XE, PGD attackResNet-50 · Tiny ImageNet PGD eps=0.04accuracy under attack (%): 0.0 -> 9.52Improved
use GCE instead of COT, BIM attackResNet-56 · CIFAR10 BIM eps=0.04accuracy under attack (%): 7.49 -> 19.59Improved

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