An intelligent forecasting system for unauthorized URL identification using deep learning

Abstract

A cyber-attack is a destructive action carried out in cyberspace by an intruder, a group, an association, or a country that targets computer information systems, infrastructures, personal computers, and servers in order to carry out harmful operations or impair their functioning. Attackers produce and sell a wide range of tools for launching web application attacks and increasing the quantity of destructive attacks. Dynamic classifiers continuously watch the web page's activities, whereas classifiers download the page, extract the attributes, and do classification analysis. This article covers how to identify a fraudulent person using a URL and other pertinent data. The process of selecting features from a URL is accomplished using PCA, and the collected characteristics are used to classify URL types. The categorization is done with 98.5 percent accuracy using the DBN approach, which distinguishes between normal and benign users. Other existing techniques are outperformed by the feature selection technique PCA with DBN classification.