AI Paper Atlas

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

Segmentations-Leak: Membership Inference Attacks and Defenses in Semantic Image Segmentation

Yang He, Shadi Rahimian, Bernt Schiele, Mario Fritz

ECCV 2020 · Adversarial robustness and security › Attacks on text and graphs · approach: Biologically inspired robust vision · 55 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 structured loss map instead of concatenated prediction and label (ablation)rejection-sampled patch attacker · Cityscapes dependent PSP->PSPattack AUC: 83.0 -> 87.1Improved
sample random patches instead of whole image (ablation)structured loss map attacker · Cityscapes dependent PSP->PSPattack AUC: 82.6 -> 84.6Improved
add rejection of uninformative (road) patches to random patch sampling (ablation)structured loss map attacker · Cityscapes dependent PSP->PSPattack AUC: 84.6 -> 87.1Improved
add rejection of uninformative patches, independent attack (ablation)structured loss map attacker, Deeplab-v3+ shadow, second victim column · Cityscapes independentattack AUC: 94.3 -> 94.9Improved
keep spatial structure (larger input blocks) in the attacker (ablation)per-patch attacker · Cityscapes dependent PSP->PSPper-patch attack performanceImproved
use 90x90 patch input instead of 1x1 vector input (ablation)concatenation attacker · Cityscapes dependentattack performanceImproved
return argmax labels only (ablation)segmentation victim model · Cityscapesattack AUCNo clear effect
add Gaussian noise to posteriors (variance 0.01 to 0.1), independent attack (ablation)PSPNet and UperNet victims · Cityscapes independentattack AUC: ~0.9 -> less than 0.6Improved
add very strong Gaussian noise (variance 0.4), dependent attack (ablation)PSPNet victim · Cityscapes dependentattack AUC and mIoU: 59.88 -> 23.17 mIoU, AUC still above 0.75Hurt
enable dropout at test time (ratios 0.1, 0.5, 0.9) (ablation)segmentation victim model · Cityscapesattack AUC and mIoUMixed
train with DPSGD (Gaussian noise on clipped gradients) (ablation)PSPNet and UperNet victims · Cityscapesattack AUC and mIoU: mIoU reduced by 1.12, 1.36 and 0.75 at noise 1e-6Improved

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