Adversarial Attacks and Defenses in Deep Learning

作     者：Kui Ren Tianhang Zheng Zhan Qin Xue Liu Kui Ren;Tianhang Zheng;Zhan Qin;Xue Liu

作者机构：Institute of Cyberspace ResearchZhejiang UniversityHangzhou 310027China College of Computer Science and TechnologyZhejiang UniversityHangzhou 310027China Department of Electrical and Computer EngineeringUniversity of TorontoTorontoON M5S 2E8Canada School of Computer ScienceMcGill UniversityMontrealQC H3A 0E9Canada 

出 版 物：《Engineering》 (工程（英文）)

年 卷 期：2020年第6卷第3期

页      面：346-360页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 0839[工学-网络空间安全] 08[工学] 0835[工学-软件工程] 081201[工学-计算机系统结构] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Ant Financial Zhejiang University Financial Technology Research Center 

主　　题：Machine learning Deep neural network Adversarial example Adversarial attack Adversarial defense 

摘      要：With the rapid developments of artificial intelligence(AI)and deep learning(DL)techniques,it is critical to ensure the security and robustness of the deployed ***,the security vulnerability of DL algorithms to adversarial samples has been widely *** fabricated samples can lead to various misbehaviors of the DL models while being perceived as benign by *** implementations of adversarial attacks in real physical-world scenarios further demonstrate their ***,adversarial attack and defense techniques have attracted increasing attention from both machine learning and security communities and have become a hot research topic in recent *** this paper,we first introduce the theoretical foundations,algorithms,and applications of adversarial attack *** then describe a few research efforts on the defense techniques,which cover the broad frontier in the *** open problems and challenges are subsequently discussed,which we hope will provoke further research efforts in this critical area.