Multiclass Classification of Malicious URL Detection Using Machine Learning and Deep Learning

Because of the rapid growth in internet usage, cyber risks have increased, especially in the form of malicious URLs. These malicious URLs are used as a means to deceive and exploit unsuspecting users, resulting in severe consequences such as data breaches, financial loss, and system compromise. The research paper presents an analysis of various techniques for malicious URL detection. The paper provides an overview on the impacts of malicious URLs on cyber attacks, phishing attempts, and malware infections, Introduces the various types of malicious URLs, and also discusses their characteristics and common strategies used by attackers to distract the malicious intent. The paper aims to detect the various malicious URLs for which URLs will be safe, risky, or too risky. Experimental results show that the random forest classifier and Feed feed-forward neural Network algorithms give better performance which is 91.44% and 95.98% respectively. The experimental results also show that the suggested URL features and behaviors can significantly improve the ability to recognize malicious URLs. This implies that the suggested methodology might be regarded as an effective method of identifying malicious URLs.