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 2007.02977cs.LGstat.ML
收藏

共享模型或核心集:基于成员推断攻击的研究

Sharing Models or Coresets: A Study based on Membership Inference Attack

Pennsylvania State University(宾夕法尼亚州立大学);IBM T. J. Watson Research Center(IBM T. J. 沃森研究中心);Army Research Laboratory(陆军研究实验室)

Hanlin Lu, Changchang Liu, Ting He, Shiqiang Wang, Kevin S. Chan

摘要:

分布式机器学习通常旨在基于分布式数据训练全局模型,而无需将所有数据收集到集中位置,其中已经提出了两种不同的方法:收集和聚合本地模型(联邦学习)以及收集和训练代表性数据摘要(coreset)。虽然由于不共享原始数据,每种方法都在一定程度上保护了数据隐私,但在试图从共享信息中推断原始数据的复杂攻击下,保护的确切程度尚不清楚。我们提出了两种方法之间的目标模型的准确性,通信成本和数据隐私方面的第一个比较,其中最后一个是衡量一个国家的最先进的攻击策略称为成员推断攻击的准确性。我们的实验量化了每种方法的准确性-隐私性-成本权衡,并揭示了一个可用于指导模型训练过程设计的重要比较。

英文摘要:

Distributed machine learning generally aims at training a global model based on distributed data without collecting all the data to a centralized location, where two different approaches have been proposed: collecting and aggregating local models (federated learning) and collecting and training over representative data summaries (coreset). While each approach preserves data privacy to some extent thanks to not sharing the raw data, the exact extent of protection is unclear under sophisticated attacks that try to infer the raw data from the shared information. We present the first comparison between the two approaches in terms of target model accuracy, communication cost, and data privacy, where the last is measured by the accuracy of a state-of-the-art attack strategy called the membership inference attack. Our experiments quantify the accuracy-privacy-cost tradeoff of each approach, and reveal a nontrivial comparison that can be used to guide the design of model training processes.

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。

↑