arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.15868cs.CLcs.LG
收藏

NLP模型的隐私评估基准

Privacy Evaluation Benchmarks for NLP Models

Ant Group(蚂蚁集团);East China Normal University(华东师范大学)

Wei Huang, Yinggui Wang, Cen Chen

AI总结:

本文提出一个覆盖传统模型与LLM的NLP隐私攻击与防御评估基准,并基于该框架研究辅助数据与攻击强度的关联,提出基于知识蒸馏的改进攻击方法及链式攻击框架。

摘要:

通过对NLP模型发起隐私攻击,攻击者可以获取训练数据和模型参数等敏感信息。尽管研究人员已经深入研究了NLP模型中的多种攻击类型,但这些分析缺乏系统性。目前缺少对攻击所造成影响的全面理解。例如,我们必须考虑哪些场景适用于哪些攻击、影响不同攻击性能的共同因素是什么、不同攻击之间关系的本质,以及各种数据集和模型对攻击有效性的影响等。因此,我们需要一个基准来整体评估NLP模型所面临的隐私风险。在本文中,我们提出了一个NLP领域的隐私攻击与防御评估基准,该基准涵盖传统/小型模型和大型语言模型(LLMs)。该基准支持多种模型、数据集和协议,并配有标准化模块,用于对攻击和防御策略进行全面评估。基于上述框架,我们研究了来自不同领域的辅助数据与隐私攻击强度之间的关联。并且借助知识蒸馏(Knowledge Distillation, KD),我们在该场景下提供了一种改进的攻击方法。此外,我们提出了一个用于隐私攻击的链式框架,允许实践者将多个攻击串联起来以实现更高层次的攻击目标。在此基础上,我们提供了一些防御和增强攻击策略。用于复现结果的代码可在https://github.com/user2311717757/nlp_doctor找到。

英文摘要:

By inducing privacy attacks on NLP models, attackers can obtain sensitive information such as training data and model parameters, etc. Although researchers have studied, in-depth, several kinds of attacks in NLP models, they are non-systematic analyses. It lacks a comprehensive understanding of the impact caused by the attacks. For example, we must consider which scenarios can apply to which attacks, what the common factors are that affect the performance of different attacks, the nature of the relationships between different attacks, and the influence of various datasets and models on the effectiveness of the attacks, etc. Therefore, we need a benchmark to holistically assess the privacy risks faced by NLP models. In this paper, we present a privacy attack and defense evaluation benchmark in the field of NLP, which includes the conventional/small models and large language models (LLMs). This benchmark supports a variety of models, datasets, and protocols, along with standardized modules for comprehensive evaluation of attacks and defense strategies. Based on the above framework, we present a study on the association between auxiliary data from different domains and the strength of privacy attacks. And we provide an improved attack method in this scenario with the help of Knowledge Distillation (KD). Furthermore, we propose a chained framework for privacy attacks. Allowing a practitioner to chain multiple attacks to achieve a higher-level attack objective. Based on this, we provide some defense and enhanced attack strategies. The code for reproducing the results can be found at https://github.com/user2311717757/nlp_doctor.

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。

↑