AI Paper Atlas

Research map / Federated learning and privacy / Privacy attacks and secure learning

Reminiscence Attack on Residuals: Exploiting Approximate Machine Unlearning for Privacy

Yaxin Xiao, Qingqing Ye, Li Hu, Huadi Zheng, Haibo Hu, Zi Liang, Haoyang Li, Yijie Jiao

ICCV 2025 · Federated learning and privacy › Privacy attacks and secure learning · approach: Membership inference and auditing · 2 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 the replay phase (two-phase design) to orthogonal unlearning (ablation)OUR · CIFAR20 ablationReA attack accuracy: ReA accuracy 95.52% without replay; unlearning efficacy degraded by 0.06Improved
use orthogonal unlearning instead of random labels in the first phase (ablation)OUR (with Replay) · CIFAR classificationReA attack accuracy: 2.89-6.12% lower ReA accuracyImproved
add l1 regularization in the replay phase (ablation)OUR · class-wise unlearningReA attack accuracy: 35.07% lower ReA accuracyImproved
add l1 regularization in the replay phase (ablation)OUR · sample-wise unlearningReA attack accuracyNo clear effect
select first, middle and last block outputs (Span 3) for orthogonal layers (ablation)OUR · CIFAR20ToW / ReA accuracyImproved
aggregate multiple learning rates in ReA instead of a single lr (ablation)ReA · CIFAR20 class-wise MIAReA attack accuracyImproved
raise ReA learning rate to 0.1 (ablation)ReA · CIFAR20 class-wise MIAReA attack accuracyHurt
set ReA convergence threshold between 70% and 80% (ablation)ReA · CIFAR20 class-wise MIAReA attack accuracy: 70% and 80%Improved
add SalUn parameter freezing to OUR (SalUn-OUR) (ablation)OUR · identity unlearning (Stable Diffusion)RISM: 0.68 -> 0.74Improved
use ReA instead of MIA-LiRA against random-label unlearningmembership inference · CIFAR20 (-5) class-wisebalanced attack accuracy: 50.00 -> 95.52Improved
apply OUR instead of RUMapproximate unlearning · CIFAR20 (-5) class-wiseReA balanced accuracy: 85.64 -> 52.28Improved
use DiffAtk instead of ReA against SalUn-unlearned SDattack · identity unlearningresonance index: ReA 2 vs DiffAtk 10 (ISM 0.77 vs 0.10)Hurt

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