Efectividad de Modelos de Machine Learning en la detección de intrusos en sistemas de información y Aplicabilidad en Emprendimiento e Innovación (#237)

Authors

Llontop Alama, Daniel Ivan

Dios-Castillo, Christian Abraham

Abstract

As technology has been advancing and the world is becoming more and more digitalized, the use of information systems in companies has become a key point for their development, and with it, security has become more relevant than ever. The purpose of this research is to explore Machine Learning models and their effectiveness in intrusion detection, evaluating their impact and applicability in the context of entrepreneurship and innovation. Therefore, 50 papers obtained from the Scopus database in the period between 2015 and 2024 focused on the use of Machine Learning models for intrusion detection were thoroughly analyzed. This review covers aspects such as: the most common types of attacks, most used dataste, most studied Machine Learning models and their classification, and finally their effectiveness in intrusion detection. The results showed that the 58 Machine Learning models identified present a minimum of 79.00% effectiveness and a maximum of 99.99% effectiveness. The conclusion is that Machine Learning models are highly effective for intrusion detection and that the implementation of these models in Intrusion Detection Systems (IDS) ensures a high percentage of the continuity of the business that is implemented online, as well as the security of the company's information and its customers.