Deep learning intelligent detection method for phishing web pages

Abstract

The invention discloses a deep learning intelligent detection method for fishing webpages, which belongs to the technical field of network information safety. The deep learning intelligent detection method comprises the following steps of (1) analyzing webpage document models to generate a webpage document feature vector F; (2) converting the webpages to be measured into images and adopting a spectral clustering method to cut the obtained images; (3) picking up characteristics of webpage images to obtain a webpage content characteristic vector N; (4) using a manifold learning Isomap algorithm to reduce dimensions of the webpage content characteristic vector N so as to obtain a characteristic space Vnew; and (5) training and testing the characteristic space Vnew by using a data base network (DBN) sorter, and judging whether the webpages to be measured are the fishing webpage or not according to results of the DBN sorter. The deep learning intelligent detection method has the advantages that measured characteristic parameter covering is comprehensive. Compared with a test characteristic extraction method, the DBN deep trust network has high detection accuracy and fast detection speed, and detection rate of fishing type attacks is improved.

Description

technical field [0001] The invention belongs to the technical field of network information security, and in particular relates to a deep learning intelligent detection method for phishing webpages. Background technique [0002] In recent years, network "phishing" attacks have appeared frequently, seriously affecting the development of e-commerce and causing great harm to the public. Common domestic "phishing" attacks (Phishing), such as phishing websites that counterfeit major banks and other financial institutions and large trading portals, are very harmful. Current detection techniques for phishing webpages are generally based solely on document models or webpage image detection methods. Due to the flexibility of the HTML language and the dynamics of webpage elements, counterfeiters can make webpages that look the same but have completely different structures, so the phishing webpage detection method based on the document model alone has great defects; similarly, the curr...

AI Technical Summary

Problems solved by technology

[0003] The present invention discloses a deep learning intelligent detection method for phishing webpages aiming at the above-mentioned defects. The method is used to solve the problem that the current phishing webpage detection technology based solely on document type or image type is not enough to process picture webpages, and the detection accuracy of phishing webpages is not high.

Images