一种基于深度学习的层次化钓鱼网站检测方法  

Hierachical Phishing Detection Method based on Deep Learning

作　　者：胡俊[1] 摆亮[1] 吕志泉[1] 

机构地区：[1]国家计算机网络应急技术处理协调中心,北京100029

出　　处：《通信技术》2017年第5期1025-1028,共4页Communications Technology

摘　　要：提出一种基于深度学习的层次化钓鱼网站检测方法,方法包括两大部分,第一部分叫"轻检测",主要对千万级的输入进行快速预判断,得出最为疑似的钓鱼网站列表,从而将原始输入规模降低到一定的数量级规模。轻检测轻便、快速、尽量准确,并达到了最少数量的漏报,从而发挥对巨大输入进行预处理和数据筛选的作用。完成第一部分的轻检测后,将筛选后的数据源输入到第二部分的重检测中。重检测是一种细粒度检测,这里采用一种基于自编码深度学习的方法,以达到将钓鱼网站从海量网站中分类出来的目的。通过两者的配合,达到系统的平衡,在保证准确性的前提下,提高整个系统的运行效率和处理能力。A hierarchical phishing detection method based on deep Learning is presented. The method involves two parts, the first part is called ＂light detection＂, which mainly makes a quick pre-judgment of the input of millions of URLs. ＂light detection＂ is light, fast, and as accurate as possible. After completing the first part of the detection, the data is transferred to the second part of the detection-＂heavy detection＂.The ＂heavy detection＂ is a method based on deep learning, for the purpose to classifying the phishing website from the mass websites. After these two steps, phishing websites would be captured. The cooperation of this two parts could improve the operation efficiency and processing ability of the entire system while ensuring detection accuravy.

关 键 词：钓鱼检测 深度学习 自编码 轻检测 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]