Web Phishing Detection Using a Deep Learning Framework

Abstract

Web service is one of the key communications software services for the Internet. Web phishing is one of many security threats to web services on the Internet. Web phishing aims to steal private information, such as usernames, passwords, and credit card details, by way of impersonating a legitimate entity. It will lead to information disclosure and property damage. This paper mainly focuses on applying a deep learning framework to detect phishing websites. This paper first designs two types of features for web phishing: original features and interaction features. A detection model based on Deep Belief Networks (DBN) is then presented. The test using real IP flows from ISP (Internet Service Provider) shows that the detecting model based on DBN can achieve an approximately 90% true positive rate and 0.6% false positive rate.

A Cognitive Support for Identifying Phishing Websites using Bi-LSTM and RNN

“…In building up the system, we utilized a portion of the broadly utilized content-based highlights, for example, breaking down Java content, source code, and word phrases (for example obsolete copyrights and "pay by telephone"). Phishing sites more often than not have many spelling and linguistic errors and long URLs [3].Another page substance highlight is the likeness score between pages' substance. Phishing sites for the most part utilizes comparable or even a similar content substance to its objective website page so as to draw their guests.…”

“…They are recognized from other neural systems by having a criticism circle associated with their past choices, and memory cells. The choice a repetitive net came to at time step − 1 influences the choice it will achieve one minute later at time step [3].…”

“…Bidirectional RNN can be prepared all the while in the positive and the negative time heading ℎ123 and ℎ143 [3]. Adding memory cells to repetitive systems empower them to remember information for long time thus comprehend the evaporating slope issue.…”

“…Phishing assaults have been depending on misdirection, redirection, and abuse of absence of client information [2]. The evaluated robbery through phishing assaults costs U.S. banks and charge card organizations $2.8 billion yearly [3]. Phishing sites are a difficult issue in light of the fact that of the expanded number of phishing sites that utilization wise techniques to hoodwink web clients.…”

“…Satire locales duplicate organization logos from the first sites and devised sites reuse pictures of items with a similar record name and size [12]. Be that as it may, picture highlights or prompts have low discovery control in examination with other extortion prompts since picture signals are extricated from picture metadata, for example, document name and measure, or from measurable highlights, for example, pixel shading frequencies [3]. One way to deal with upgrade picture signals execution is to build up a device that can perceive items and activities inside pictures.…”