On the Effectiveness of Membership Inference in Targeted Data Extraction from Large Language Models
American University of Beirut(贝鲁特美国大学)
Ali Al Sahili, Ali Chehab, Razane Tajeddine
备注:
This work has been accepted for publication at the IEEE Conference on Secure and Trustworthy Machine Learning (SaTML). The final version will be available on IEEE Xplore
Large Language Models (LLMs) are prone to memorizing training data, which poses serious privacy risks. Two of the most prominent concerns are training data extraction and Membership Inference Attacks (MIAs). Prior research has shown that these threats are interconnected: adversaries can extract training data from an LLM by querying the model to generate a large volume of text and subsequently applying MIAs to verify whether a particular data point was included in the training set. In this study, we integrate multiple MIA techniques into the data extraction pipeline to systematically benchmark their effectiveness. We then compare their performance in this integrated setting against results from conventional MIA benchmarks, allowing us to evaluate their practical utility in real-world extraction scenarios.
Ethicist: Targeted Training Data Extraction Through Loss Smoothed Soft Prompting and Calibrated Confidence Estimation
Institute for Artificial Intelligence(人工智能研究院);Beijing National Research Center for Information Science and Technology(北京信息科学与技术国家研究中心);Tsinghua University(清华大学)
我们研究了自然语言处理中的模型提取问题,在这个问题中,一个只对受害者模型进行查询访问的对手试图重建该模型的本地副本。假设对手和受害者模型都对大型预训练语言模型进行了微调,如BERT(Devlin et al. 2019),我们证明对手不需要任何真实的训练数据来成功发起攻击。事实上,攻击者甚至不需要使用语法或语义上有意义的查询:我们表明,随机的单词序列加上任务特定的语法,形成有效的查询模型提取不同的NLP任务,包括自然语言推理和问答。因此,我们的工作突出了一个只有在NLP社区转向迁移学习方法时才变得可行的漏洞:对于几百美元的查询预算,攻击者可以提取一个性能仅略差于受害者模型的模型。最后,我们研究了两种防御策略对模型提取-成员分类和API水印-而成功的对天真的对手,是无效的对更复杂的。
Auditing Data Membership in Reinforcement Learning With Verifiable Rewards
The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州));Shanghai Jiao Tong University(上海交通大学);Shandong University(山东大学);Xidian University(西安电子科技大学);Wuhan University(武汉大学)
MeasHalu: Mitigation of Scientific Measurement Hallucinations for Large Language Models with Enhanced Reasoning
Shenzhen Key Laboratory for High Performance Data Mining(深圳高性能数据挖掘重点实验室);Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences(深圳先进技术研究院,中国科学院);Artificial Intelligence Research Institute, Shenzhen University of Advanced Technology(深圳先进技术大学人工智能研究院)
An Exact Poly-Time Membership-Queries Algorithm for Extraction a three-Layer ReLU Network
The Hebrew University(希伯来大学);Google Research(谷歌研究院)
我们考虑了从查询中学习ReLU网络的自然问题,该问题最近被模型窃取攻击重新激活。在这项工作中,我们提出了一个多项式时间算法,可以在温和的一般位置假设下从查询中学习深度为2的ReLU网络。我们还提出了一个多项式时间算法,在温和的一般位置假设下,可以从查询中学习丰富的深度为3的ReLU网络。例如,它可以学习大多数第一层神经元的数量小于第二层神经元的维数和数量的网络。这两个结果大大提高了最新技术水平:在我们的工作之前,多项式时间算法仅在假设基础分布为高斯分布(Chen et al.(2021))或权重矩阵行线性独立(Milli et al.(2019))的情况下从查询深度为2的网络中学习。对于深度3或以上,没有已知的多时间结果。
AI-PAVE-Br: Leveraging Large Language Models for Enhanced Product Attribute Value Extraction through a Golden Set Approach
LuizaLabs – Center of Excellence in Artificial Intelligence(LuizaLabs人工智能卓越中心);Department of Computing and Informatics – Mackenzie Presbyterian University(计算与信息学系-麦金西 Presbyterian大学);Institute of Mathematics and Computer Sciences - University of São Paulo(数学与计算机科学研究所-圣保罗大学)
A Multi-Stage Validation Framework for Trustworthy Large-scale Clinical Information Extraction using Large Language Models
Oak Ridge National Laboratory(橡树岭国家实验室);Program Evaluation and Resource Center, Office of Mental Health and Office of Suicide Prevention, Department of Veterans Affairs(退伍军人事务部心理健康办公室和自杀预防办公室项目评估与资源中心);Vanderbilt University Medical Center(范德比尔特大学医学中心);VA Maryland Health Care System(退伍军人事务部马里兰医疗保健系统);VA Desert Pacific Healthcare Network(退伍军人事务部沙漠太平洋医疗网络);VA Connecticut Health Care System(退伍军人事务部康涅狄格医疗保健系统);Yale School of Medicine(耶鲁医学院)
A Survey on Model Extraction Attacks and Defenses for Large Language Models
University of Notre Dame(圣约翰大学);Florida State University(佛罗里达州立大学);Northwestern University(西北大学);Duke University(杜克大学);University of Southern California(南加州大学)
Targeted Tests for LLM Reasoning: An Audit-Constrained Protocol
School of Life Science and Technology, Institute of Science Tokyo(生命科学与技术学院,科学东京研究所);Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences(计算生物学与医学科学系,前沿科学研究生院)
Agentic Literature Extraction Augments Targeted Protein Degradation Databases
Department of Computer Science and Engineering(计算机科学与工程系);Chalmers University of Technology(楚姆勒斯技术大学);University of Gothenburg(哥德堡大学);Yardi School of Artificial Intelligence(Yardi人工智能学院);Indian Institute of Technology Delhi(德里印度理工学院)
Marich: A Query-efficient Distributionally Equivalent Model Extraction Attack using Public Data
Ramakrishna Mission Vivekananda Educational and Research Institute(罗摩克里希纳传教会维韦卡南达教育与研究院);National University of Singapore(新加坡国立大学);Univ. Lille(里尔大学);Inria(法国国家信息与自动化研究所);CNRS(法国国家科学研究中心);Centrale Lille(里尔中央理工学院)
我们研究黑盒模型提取攻击的设计,这些攻击通过预测 API 向目标ML模型发送极少查询。旨在从公开可用的数据集中提取一个目标ML模型,旨在旨在创建一个信息丰富、分布等_equ价的副本
Research on the pixel-based and object-oriented methods of urban feature extraction with GF-2 remote-sensing images
East China Normal University(华东师范大学);Peter the Great St. Petersburg Polytechnic University(彼得大帝圣彼得堡理工大学);Institute of Technical Physics, Chinese Academy of Sciences(中国科学院上海技术物理研究所)