基于侧信道攻击的深度神经网络黑盒对抗攻击的数据还原

IF 4.8 2区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS
Hanxun Zhou , Zhihui Liu , Yufeng Hu , Shuo Zhang , Longyu Kang , Yong Feng , Yan Wang , Wei Guo , Cliff C. Zou
{"title":"基于侧信道攻击的深度神经网络黑盒对抗攻击的数据还原","authors":"Hanxun Zhou ,&nbsp;Zhihui Liu ,&nbsp;Yufeng Hu ,&nbsp;Shuo Zhang ,&nbsp;Longyu Kang ,&nbsp;Yong Feng ,&nbsp;Yan Wang ,&nbsp;Wei Guo ,&nbsp;Cliff C. Zou","doi":"10.1016/j.cose.2025.104401","DOIUrl":null,"url":null,"abstract":"<div><div>Launching effective black-box adversarial attack against a deep neural network (DNN) without knowledge of the model's details is challenging. Previous studies involved performing numerous queries on the target model to generate adversarial examples, which is unacceptable due to the high query volume. Additionally, many of these queries are unnecessary as the dataset may contain redundant or duplicate data. To address these issues, we propose a two-stage black-box adversarial attack approach that combines side-channel attacks and a data reduction technique. In the first stage, we employ Long Short Term Memory (LSTM) to gather partial information about the target DNN through side-channel attacks, enabling us to obtain the class probability of the dataset. In the second stage, we utilize a new data reduction algorithm based on the class probability to enhance the efficiency of generating adversarial examples. Our approach is capable of precisely identifying the target model and the data reduction performs better than other reduction methods. Furthermore, when utilizing the reduced datasets to train the shadow model, the adversarial examples generated on this shadow model demonstrate a higher transferability success rate than SOTA data reduction methods.</div></div>","PeriodicalId":51004,"journal":{"name":"Computers & Security","volume":"153 ","pages":"Article 104401"},"PeriodicalIF":4.8000,"publicationDate":"2025-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Data reduction for black-box adversarial attacks against deep neural networks based on side-channel attacks\",\"authors\":\"Hanxun Zhou ,&nbsp;Zhihui Liu ,&nbsp;Yufeng Hu ,&nbsp;Shuo Zhang ,&nbsp;Longyu Kang ,&nbsp;Yong Feng ,&nbsp;Yan Wang ,&nbsp;Wei Guo ,&nbsp;Cliff C. Zou\",\"doi\":\"10.1016/j.cose.2025.104401\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Launching effective black-box adversarial attack against a deep neural network (DNN) without knowledge of the model's details is challenging. Previous studies involved performing numerous queries on the target model to generate adversarial examples, which is unacceptable due to the high query volume. Additionally, many of these queries are unnecessary as the dataset may contain redundant or duplicate data. To address these issues, we propose a two-stage black-box adversarial attack approach that combines side-channel attacks and a data reduction technique. In the first stage, we employ Long Short Term Memory (LSTM) to gather partial information about the target DNN through side-channel attacks, enabling us to obtain the class probability of the dataset. In the second stage, we utilize a new data reduction algorithm based on the class probability to enhance the efficiency of generating adversarial examples. Our approach is capable of precisely identifying the target model and the data reduction performs better than other reduction methods. Furthermore, when utilizing the reduced datasets to train the shadow model, the adversarial examples generated on this shadow model demonstrate a higher transferability success rate than SOTA data reduction methods.</div></div>\",\"PeriodicalId\":51004,\"journal\":{\"name\":\"Computers & Security\",\"volume\":\"153 \",\"pages\":\"Article 104401\"},\"PeriodicalIF\":4.8000,\"publicationDate\":\"2025-02-25\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Computers & Security\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0167404825000902\",\"RegionNum\":2,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"COMPUTER SCIENCE, INFORMATION SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computers & Security","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0167404825000902","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
引用次数: 0

摘要

在不了解模型细节的情况下对深度神经网络(DNN)发起有效的黑盒对抗性攻击是具有挑战性的。以前的研究涉及在目标模型上执行大量查询以生成对抗性示例,由于高查询量,这是不可接受的。此外,许多这些查询是不必要的,因为数据集可能包含冗余或重复的数据。为了解决这些问题,我们提出了一种结合了侧信道攻击和数据约简技术的两阶段黑箱对抗性攻击方法。在第一阶段,我们使用长短期记忆(LSTM)通过侧信道攻击收集目标DNN的部分信息,使我们能够获得数据集的类概率。在第二阶段,我们利用一种新的基于类概率的数据约简算法来提高生成对抗样本的效率。我们的方法能够精确地识别目标模型,并且数据约简的性能优于其他约简方法。此外,当利用约简后的数据集训练阴影模型时,在该阴影模型上生成的对抗样例显示出比SOTA数据约简方法更高的可转移成功率。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Data reduction for black-box adversarial attacks against deep neural networks based on side-channel attacks
Launching effective black-box adversarial attack against a deep neural network (DNN) without knowledge of the model's details is challenging. Previous studies involved performing numerous queries on the target model to generate adversarial examples, which is unacceptable due to the high query volume. Additionally, many of these queries are unnecessary as the dataset may contain redundant or duplicate data. To address these issues, we propose a two-stage black-box adversarial attack approach that combines side-channel attacks and a data reduction technique. In the first stage, we employ Long Short Term Memory (LSTM) to gather partial information about the target DNN through side-channel attacks, enabling us to obtain the class probability of the dataset. In the second stage, we utilize a new data reduction algorithm based on the class probability to enhance the efficiency of generating adversarial examples. Our approach is capable of precisely identifying the target model and the data reduction performs better than other reduction methods. Furthermore, when utilizing the reduced datasets to train the shadow model, the adversarial examples generated on this shadow model demonstrate a higher transferability success rate than SOTA data reduction methods.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Computers & Security
Computers & Security 工程技术-计算机:信息系统
CiteScore
12.40
自引率
7.10%
发文量
365
审稿时长
10.7 months
期刊介绍: Computers & Security is the most respected technical journal in the IT security field. With its high-profile editorial board and informative regular features and columns, the journal is essential reading for IT security professionals around the world. Computers & Security provides you with a unique blend of leading edge research and sound practical management advice. It is aimed at the professional involved with computer security, audit, control and data integrity in all sectors - industry, commerce and academia. Recognized worldwide as THE primary source of reference for applied research and technical expertise it is your first step to fully secure systems.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信