Noise Aggregation Analysis Driven by Small-Noise Injection: Efficient Membership Inference for Diffusion Models

TL;DR

This paper introduces a novel, efficient membership inference attack on diffusion models using noise aggregation analysis and a low-noise injection strategy to improve accuracy and reduce computational costs.

Contribution

It proposes a new noise aggregation analysis method combined with a single-step noise injection technique to enhance membership inference on diffusion models.

Findings

Achieves higher inference accuracy with fewer model queries.

Reduces computational costs compared to existing methods.

Effectively amplifies differences between member and non-member samples.

Abstract

Diffusion models have demonstrated powerful performance in generating high-quality images. A typical example is text-to-image generator like Stable Diffusion. However, their widespread use also poses potential privacy risks. A key concern is membership inference attacks, which attempt to determine whether a particular data sample was used in the model training process. Existing membership inference attacks against diffusion models either directly exploit sample loss differences or rely on image-level reconstruction differences. Both approaches commonly ignore the consistency characteristics of noise prediction during the diffusion process, resulting in either low inference accuracy or high computational costs. To address these shortcomings, we propose a membership inference method based on noise aggregation analysis, and introduce a single-step, low-intensity noise injection diffusion strategy to amplify differences between member and non-member samples. Our proposed approach substantially reduces model query requirements while delivering more efficient and accurate membership inference.