Imitative Membership Inference Attack

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Du, Yuntao, Chen, Yuetian, Xiao, Hanshen, Ribeiro, Bruno, Li, Ninghui
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914244016996352
author Du, Yuntao
Chen, Yuetian
Xiao, Hanshen
Ribeiro, Bruno
Li, Ninghui
author_facet Du, Yuntao
Chen, Yuetian
Xiao, Hanshen
Ribeiro, Bruno
Li, Ninghui
contents A Membership Inference Attack (MIA) assesses how much a target machine learning model reveals about its training data by determining whether specific query instances were part of the training set. State-of-the-art MIAs rely on training hundreds of shadow models that are independent of the target model, leading to significant computational overhead. In this paper, we introduce Imitative Membership Inference Attack (IMIA), which employs a novel imitative training technique to strategically construct a small number of target-informed imitative models that closely replicate the target model's behavior for inference. Extensive experimental results demonstrate that IMIA substantially outperforms existing MIAs in various attack settings while only requiring less than 5% of the computational cost of state-of-the-art approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2509_06796
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Imitative Membership Inference Attack
Du, Yuntao
Chen, Yuetian
Xiao, Hanshen
Ribeiro, Bruno
Li, Ninghui
Cryptography and Security
Machine Learning
A Membership Inference Attack (MIA) assesses how much a target machine learning model reveals about its training data by determining whether specific query instances were part of the training set. State-of-the-art MIAs rely on training hundreds of shadow models that are independent of the target model, leading to significant computational overhead. In this paper, we introduce Imitative Membership Inference Attack (IMIA), which employs a novel imitative training technique to strategically construct a small number of target-informed imitative models that closely replicate the target model's behavior for inference. Extensive experimental results demonstrate that IMIA substantially outperforms existing MIAs in various attack settings while only requiring less than 5% of the computational cost of state-of-the-art approaches.
title Imitative Membership Inference Attack
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2509.06796