arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2205.02130cs.CRcs.CLcs.LG
收藏

字级差分隐私的局限性

The Limits of Word Level Differential Privacy

RWTH Aachen University(亚琛工业大学);SAP Security Research(SAP安全研究院);University of Waterloo(滑铁卢大学)

Justus Mattern, Benjamin Weggenmann, Florian Kerschbaum

摘要:

随着隐私和信任问题在研究界越来越受到关注,已经做出了各种尝试来匿名化文本数据。这些方法中的一个重要子集结合了不同的私有机制来干扰词嵌入,从而替换句子中的单个词。虽然这些方法代表了非常重要的贡献,与其他技术相比具有各种优势,并且确实显示出匿名化能力,但它们有几个缺点。在本文中,我们调查这些弱点,并展示了显着的数学约束,减少理论上的隐私保证,以及主要的实际缺陷方面的保护,防止去匿名攻击,保留原始句子的内容,以及语言输出的质量。最后,我们提出了一种新的文本匿名化方法的基础上Transformer为基础的语言模型微调的释义,规避了大多数已确定的弱点,也提供了一个正式的隐私保证。我们评估我们的方法的性能,通过彻底的实验,并表现出优越的性能超过所讨论的机制。

英文摘要:

As the issues of privacy and trust are receiving increasing attention within the research community, various attempts have been made to anonymize textual data. A significant subset of these approaches incorporate differentially private mechanisms to perturb word embeddings, thus replacing individual words in a sentence. While these methods represent very important contributions, have various advantages over other techniques and do show anonymization capabilities, they have several shortcomings. In this paper, we investigate these weaknesses and demonstrate significant mathematical constraints diminishing the theoretical privacy guarantee as well as major practical shortcomings with regard to the protection against deanonymization attacks, the preservation of content of the original sentences as well as the quality of the language output. Finally, we propose a new method for text anonymization based on transformer based language models fine-tuned for paraphrasing that circumvents most of the identified weaknesses and also offers a formal privacy guarantee. We evaluate the performance of our method via thorough experimentation and demonstrate superior performance over the discussed mechanisms.

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。

↑