# Rectified Flows 沿插值路径的成员信号泄露分析

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-06-05 21:46
- AIHOT 分数：54
- AIHOT 链接：https://aihot.virxact.com/items/cmq6pd9ed0ajvsl5idmaarv02
- 原文链接：https://arxiv.org/abs/2606.07271

## AI 摘要

研究分析 Rectified Flows 生成模型在插值路径 X_λ = (1-λ)X_0 + λX_1 上的训练数据成员信息泄露。训练集与测试集的重建误差在 λ 轴上呈钟形曲线，该差距随训练累积，而验证指标保持稳定。钟形峰值在 Gaussian 假设下有闭合解析解，并在音频与图像数据上验证其普适性。利用该 λ 分辨结构可实现成员推断攻击（MIA），区分训练集与非训练集样本。

## 正文

Understanding what generative models retain from training data remains challenging, with implications for copyright and privacy. Beyond verbatim reproduction, models can encode subtler traces of their training data that never surface in their outputs yet remain exploitable. We study this regime for Rectified Flows, which are increasingly used in deployed generative systems. We analyse the interpolation path X_λ= (1-λ)X_0 + λX_1 that defines the Rectified Flow training. We show that a gap exists between the reconstruction of train and test data that follows a bell-shaped curve over λ, wich accumulates during training, while the validation metrics remain stable. The signal has a maximum whose location we derive in closed form under Gaussian assumptions. We validate these predictions on both audio and images and show that the bell-shaped structure is universal, while the peak prediction holds when our assumptions are satisfied. As a proof of concept, we exploit this specific λ-resolved structure to perform a Membership Inference Attack, distinguishing members of the training set from non-members.