arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1902.04416cs.CRcs.AI
收藏

基于图的IoT恶意软件检测系统的对抗性学习研究

Examining Adversarial Learning against Graph-based IoT Malware Detection Systems

University of Central Florida(中佛罗里达大学)

Ahmed Abusnaina, Aminollah Khormali, Hisham Alasmary, Jeman Park, Afsah Anwar, Ulku Meteriz, Aziz Mohaisen

摘要:

本研究的主要目标是研究基于图的深度学习(DL)模型用于物联网(IoT)恶意软件分类对抗对抗性学习(AL)的鲁棒性。我们设计了两种方法来制作对抗物联网软件,包括现成的对抗攻击(OSAA)方法,使用六种不同的AL攻击方法,以及图形嵌入和增强(GEA)。GEA方法旨在通过将良性样本小心嵌入到恶意样本中来保留生成的对抗样本的功能和实用性。我们的评估表明,OSAAs能够实现100%的错误分类率(MR)。此外,我们观察到GEA方法能够将所有物联网恶意软件样本错误分类为良性。

英文摘要:

The main goal of this study is to investigate the robustness of graph-based Deep Learning (DL) models used for Internet of Things (IoT) malware classification against Adversarial Learning (AL). We designed two approaches to craft adversarial IoT software, including Off-the-Shelf Adversarial Attack (OSAA) methods, using six different AL attack approaches, and Graph Embedding and Augmentation (GEA). The GEA approach aims to preserve the functionality and practicality of the generated adversarial sample through a careful embedding of a benign sample to a malicious one. Our evaluations demonstrate that OSAAs are able to achieve a misclassification rate (MR) of 100%. Moreover, we observed that the GEA approach is able to misclassify all IoT malware samples as benign.

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。

↑