The Tail Tells All: Estimating Model-Level Membership Inference Vulnerability Without Reference Models

TL;DR

This paper introduces a new, reference-model-free method to estimate the privacy vulnerability of AI models against membership inference attacks by analyzing loss distribution characteristics.

Contribution

It proposes a novel approach to assess model vulnerability without expensive reference models, using loss distribution asymmetry and outlier absence as indicators.

Findings

Accurately estimates vulnerability to state-of-the-art MIAs

Outperforms low-cost reference-based attacks like RMIA

Effective for large-language models

Abstract

Membership inference attacks (MIAs) have emerged as the standard tool for evaluating the privacy risks of AI models. However, state-of-the-art attacks require training numerous, often computationally expensive, reference models, limiting their practicality. We present a novel approach for estimating model-level vulnerability, the TPR at low FPR, to membership inference attacks without requiring reference models. Empirical analysis shows loss distributions to be asymmetric and heavy-tailed and suggests that most points at risk from MIAs have moved from the tail (high-loss region) to the head (low-loss region) of the distribution after training. We leverage this insight to propose a method to estimate model-level vulnerability from the training and testing distribution alone: using the absence of outliers from the high-loss region as a predictor of the risk. We evaluate our method, the TNR of a simple loss attack, across a wide range of architectures and datasets and show it to accurately estimate model-level vulnerability to the SOTA MIA attack (LiRA). We also show our method to outperform both low-cost (few reference models) attacks such as RMIA and other measures of distribution difference. We finally evaluate the use of non-linear functions to evaluate risk and show the approach to be promising to evaluate the risk in large-language models.