AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Knowledge-Enriched Distributional Model Inversion Attacks

Si Chen, Mostafa Kahla, Ruoxi Jia, Guo-Jun Qi

ICCV 2021 · Adversarial robustness and security › Attacks on text and graphs · approach: Membership inference and inversion · 114 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 soft-label discrimination (SD) (ablation)GMI · CelebA VGG16attack accuracy: .21 -> .35Improved
add soft-label discrimination (SD) (ablation)GMI · CelebA VGG16FID: 52.51 -> 33.75Improved
add entropy minimization (EM) to SD (ablation)SD · CelebA VGG16attack accuracy: .35 -> .43Improved
add distributional recovery (DR) (ablation)GMI · CelebA VGG16attack accuracy: .21 -> .47Improved
add distributional recovery (DR) (ablation)GMI · CelebA VGG16FID: 52.51 -> 45.28Improved
add DR to SD (ablation)SD · CelebA VGG16attack accuracy: .35 -> .62Improved
add EM to SD+DR (ablation)SD+DR · CelebA VGG16attack accuracy: .62 -> .72Improved
add EM to SD+DR (ablation)SD+DR · CelebA VGG16FID: 23.82 -> 23.72No clear effect
replace GMI with proposed attackface.evolve · CelebAattack accuracy: .31 -> .81Improved
replace GMI with proposed attackMNIST digit classifier · MNISTattack accuracy: .08 -> .68Improved
train GAN on a shifted public dataset (FaceScrub instead of in-distribution)Ours · FaceScrub->CelebAattack accuracy: .72 -> .13Hurt
train GAN on FFHQ instead of CelebA-disjoint public dataOurs · FFHQ->CelebAattack accuracy: .72 -> .36Hurt

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