跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.25640 · v1

Erasing the Wrong Thing: Conditional Attribute Leakage and Concept Erasure in Shared Foundation-Model Embeddings

首次提交 2026-11-19 15:27
版本更新 2026-11-19 15:27
雷达观测 2026-11-20 19:58
arXiv ↗PDF ↗

摘要

Medical-image embeddings are increasingly shared in place of raw images for authorized downstream tasks, yet they may reveal sensitive attributes. Standard concept-erasure methods target population-level guardedness. However, we prove that, when binary task and sensitive labels are associated, global and task-conditional linear guardedness cannot both hold while a nonzero linear task signal is preserved, so global erasure must either eliminate that signal or retain task-conditional linear leakage. We further show how global erasure affects linear readers' discrimination between positive and negative samples with different sensitive attributes, deriving an approximate closed-form prediction of its utility cost from pre-erasure statistics. Motivated by this conflict, we introduce Task-Conditional LEACE (TC-LEACE), a closed-form variant of LEACE that enforces linear guardedness within each task stratum of its fitting data. We evaluate four vision foundation models on 85 CelebA attribute pairs spanning a broad range of associations. After LEACE, conditional leakage persists across all four encoders, the discrepancy between marginal and conditional audits increases with association strength, and predicted utility costs track observed losses. We then examine hospital chest radiographs using three foundation models, including two biomedical models, in which chest-tube status is strongly associated with diagnosis. Although a marginal attack on LEACE-erased embeddings underperforms a diagnosis-only reference, chest-tube status remains detectable within diagnosis strata. TC-LEACE recovers most of LEACE's linear task-utility loss while keeping task-conditional leakage within a prespecified margin of LEACE's.

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

目前只收录了一个版本。收录下一版后,可以对照标题、作者、分类和摘要。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v1 · 2026-11-19 15:27 · 最新观测版本

Erasing the Wrong Thing: Conditional Attribute Leakage and Concept Erasure in Shared Foundation-Model Embeddings

Medical-image embeddings are increasingly shared in place of raw images for authorized downstream tasks, yet they may reveal sensitive attributes. Standard concept-erasure methods target population-level guardedness. However, we prove that, when binary task and sensitive labels are associated, global and task-conditional linear guardedness cannot both hold while a nonzero linear task signal is preserved, so global erasure must either eliminate that signal or retain task-conditional linear leakage. We further show how global erasure affects linear readers' discrimination between positive and negative samples with different sensitive attributes, deriving an approximate closed-form prediction of its utility cost from pre-erasure statistics. Motivated by this conflict, we introduce Task-Conditional LEACE (TC-LEACE), a closed-form variant of LEACE that enforces linear guardedness within each task stratum of its fitting data. We evaluate four vision foundation models on 85 CelebA attribute pairs spanning a broad range of associations. After LEACE, conditional leakage persists across all four encoders, the discrepancy between marginal and conditional audits increases with association strength, and predicted utility costs track observed losses. We then examine hospital chest radiographs using three foundation models, including two biomedical models, in which chest-tube status is strongly associated with diagnosis. Although a marginal attack on LEACE-erased embeddings underperforms a diagnosis-only reference, chest-tube status remains detectable within diagnosis strata. TC-LEACE recovers most of LEACE's linear task-utility loss while keeping task-conditional leakage within a prespecified margin of LEACE's.

查看此版本解读 →查看此版本原文 ↗