典型文献
Kernel-based adversarial attacks and defenses on support vector classification
文献摘要：
While malicious samples are widely found in many application fields of machine learning,suitable counter-measures have been investigated in the field of adversarial machine learning.Due to the importance and popu-larity of Support Vector Machines(SVMs),we first describe the evasion attack against SVM classification and then propose a defense strategy in this paper.The evasion attack utilizes the classification surface of SVM to iteratively find the minimal perturbations that mislead the nonlinear classifier.Specially,we propose what is called a vulnerability function to measure the vulnerability of the SVM classifiers.Utilizing this vulnerability function,we put forward an effective defense strategy based on the kernel optimization of SVMs with Gaussian kernel against the evasion attack.Our defense method is verified to be very effective on the benchmark datasets,and the SVM classifier becomes more robust after using our kernel optimization scheme.
文献关键词：
中图分类号：
作者姓名：
Wanman Li;Xiaozhang Liu;Anli Yan;Jie Yang
作者机构：
School of Computer Science and Technology,Hainan University,Haikou,570 228,China
文献出处：
引用格式：
[1]Wanman Li;Xiaozhang Liu;Anli Yan;Jie Yang-.Kernel-based adversarial attacks and defenses on support vector classification)[J].数字通信与网络（英文）,2022(04):492-497
A类：
mislead
B类：
Kernel,adversarial,attacks,defenses,support,vector,classification,While,malicious,samples,are,widely,found,many,application,fields,machine,learning,suitable,counter,measures,have,been,investigated,Due,importance,popu,larity,Support,Vector,Machines,SVMs,we,first,describe,evasion,against,then,propose,strategy,this,paper,utilizes,surface,iteratively,find,minimal,perturbations,that,nonlinear,Specially,what,called,vulnerability,function,classifiers,Utilizing,put,forward,effective,kernel,optimization,Gaussian,Our,method,verified,very,benchmark,datasets,becomes,more,robust,after,using,our,scheme
AB值：
0.577403
相似文献
机标中图分类号，由域田数据科技根据网络公开资料自动分析生成，仅供学习研究参考。
