AI Paper Atlas

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

Class-Disentanglement and Applications in Adversarial Detection and Defense

Kaiwen Yang, Tianyi Zhou, Yonggang Zhang, Xinmei Tian, Dacheng Tao

NeurIPS 2021 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training with feature consistency · 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
run detectors (KD, LID, MD) on class-essential part R(x)=x-G(x) instead of xKD detector · Restricted ImageNet FGSMTNR: 8.76 -> 99.40Improved
run MD detector on R(x) instead of xMD detector · CIFAR-10 PGD-l2TNR: 76.70 -> 99.13Improved
run LID detector on R(x) instead of xLID detector · CIFAR-10 C&WTNR: 74.98 -> 76.42Improved
run LID detector on R(x) instead of xLID detector · Restricted ImageNet BIMTNR: 97.92 -> 99.46Mixed
classify G(x) from CD-VAE instead of x (grey-box defense, no adversarial training data)CD-VAE defense · CIFAR-10 PGD-l-infdefense accuracy (%): 0.0 -> 77.05Improved
CD-VAE G(x) defense vs adversarial training on StAdv attackCD-VAE defense · CIFAR-10 StAdvdefense accuracy (%): 53.9 -> 19.41Hurt
CD-VAE G(x) defense costs clean accuracyCD-VAE defense · CIFAR-10 cleanaccuracy (%): 96.01 -> 86.81Hurt
white-box adversarial training that pushes distorted class information into R(x')CD-VAE · CIFAR-10 unseen attacksmean defense accuracy (%): 50.3 -> 51.4Improved
increase gamma (classification weight) in CD-VAE objective (ablation)CD-VAE · CIFAR-10, Restricted ImageNetdetection and defense accuracyMixed
classifier trained on G(x) test on x or R(x) (information in G(x) is separate from R(x)) (ablation)classifiers on CD-VAE parts · CIFAR-10top-1 accuracy (%): 96.20 -> 25.67Mixed

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