AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

How does Bayesian Sampling help Membership Inference Attacks?

Zhenlong Liu, Wenyu Jiang, Feng Zhou, Hongxin Wei

ICML 2026 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Nonparametric and PAC-Bayes Bayesian learning

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 Laplace approximation with MC Dropout (ablation)BMIA · CIFAR-10TPR@1%FPR (%): 9.48 -> 7.94Hurt
replace Laplace approximation with SWAG (ablation)BMIA · CIFAR-10TPR@1%FPR (%): 9.48 -> 6.61Hurt
replace Laplace approximation with BatchEnsemble (ablation)BMIA · CIFAR-10TPR@1%FPR (%): 9.48 -> 1.41Hurt
increase number of posterior samples M (ablation)BMIA · CIFAR-10/100TPRImproved
use KFAC instead of diagonal Hessian factorization (ablation)BMIA · CIFAR-10/100TPRNo clear effect
use Bayesian sampling on one reference model instead of RMIA with one reference modelRMIA · CIFAR-100TPR@1%FPR (%): 10.08 -> 35.75Improved
use Bayesian sampling on one reference model instead of QMIAQMIA · CIFAR-100TPR@1%FPR (%): 15.26 -> 35.75Improved
use BMIA with 1 reference model instead of LiRA with 8 reference modelsLiRA (n=8) · CIFAR-100TPR@1%FPR (%): 23.20 -> 35.75Improved
use BMIA with 1 reference model instead of LiRA with 8 reference modelsLiRA (n=8) · Purchase-100TPR@1%FPR (%): 4.60 -> 4.61No clear effect
use BMIA with 1 reference model instead of RMIA with 8 reference modelsRMIA (n=8) · ImageNetTPR@0.1%FPR (%): 4.39 -> 4.39No clear effect
use multi-reference MR-BMIA with 64 reference modelsLiRA · CIFAR-100TPR@1%FPR (%): 43.33 -> 45.57Improved
increase reference models for LiRA from n=2 to n=8LiRA · CIFAR-100TPR@1%FPR (%): 3.67 -> 23.20Mixed

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