arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2405.04032cs.CRcs.AI
收藏

局部差分隐私的上下文学习

Locally Differentially Private In-Context Learning

Renmin University of China(中国人民大学)

Chunyan Zheng, Keke Sun, Wenhao Zhao, Haibo Zhou, Lixin Jiang, Shaoyang Song, Chunlai Zhou

AI总结:

针对 LLM 提示中敏感标签可能遭受 MIA 与提示泄露攻击的问题,提出 LDP-ICL 框架,并分析分类任务中的隐私—效用权衡及离散分布估计应用。

摘要:

大型预训练语言模型(LLM)已展现出令人惊讶的上下文学习(ICL)能力。部署大型语言模型的一个重要应用是利用私有数据库增强 LLM,以完成某些特定任务。这一颇具前景的商业用途所面临的主要问题是:LLM 已被证明会记忆其训练数据,并且其提示数据容易受到成员推理攻击(MIA)和提示泄露攻击。为了解决这一问题,我们在隐私层面将 LLM 视为不可信,并在标签敏感的设置下提出了一种用于上下文学习的局部差分隐私框架(LDP-ICL)。考虑到 Transformer 中通过梯度下降实现上下文学习的机制,我们分析了此类用于分类的 LDP-ICL 在隐私与效用之间的权衡。此外,我们将 LDP-ICL 应用于离散分布估计问题。最后,我们进行了若干实验以证明我们的分析结果。

英文摘要:

Large pretrained language models (LLMs) have shown surprising In-Context Learning (ICL) ability. An important application in deploying large language models is to augment LLMs with a private database for some specific task. The main problem with this promising commercial use is that LLMs have been shown to memorize their training data and their prompt data are vulnerable to membership inference attacks (MIA) and prompt leaking attacks. In order to deal with this problem, we treat LLMs as untrusted in privacy and propose a locally differentially private framework of in-context learning(LDP-ICL) in the settings where labels are sensitive. Considering the mechanisms of in-context learning in Transformers by gradient descent, we provide an analysis of the trade-off between privacy and utility in such LDP-ICL for classification. Moreover, we apply LDP-ICL to the discrete distribution estimation problem. In the end, we perform several experiments to demonstrate our analysis results.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑