A Study on: Implementation of Machine Learning in Malicious Emails

the main aim of the current study has been to provide a novel tool for detecting phishing attacks and finding a solution to counteract such types of threats. In this article we describe the process of how to develop a Scrum-based implementation of algorithms for automatic learning, Feature Selection and Neural Networks. This tool has the ability to detect and mitigate a phishing attack registered inside the e-mail server. For the validation of the obtained results, we have used the source of information of blacklist of PhishTank, which is a collaborative clearing house for data and information about phishing on the Internet. The conducted proof of concept demonstrated that the implemented feature selection algorithm discards the irrelevant characteristics of electronic mail and, that the neural network algorithm adopts these characteristics, establishing an optimal level of learning without redundancies. It also reveals the functionality of the proposed solution.