Detecção de Intrusão em Redes de Computadores Usando Técnicas de Detecção de Outliers

Abstract

Intrusion detection in computer networks is an area of utmost importance to ensure information security and system integrity. In this work, the use of unsupervised learning techniques for this purpose was explored. The unsupervised approach is particularly interesting because it does not require a labeled dataset for training, making it more flexible and adaptable to new forms of threats. This is crucial given the constant evolution of cyber threats and the need for dynamic and adaptable security systems. The study focuses on comparatively evaluating the performance of different algorithms in intrusion detection. Evaluation metrics such as precision, recall, and F1-score were used to compare the performance of each algorithm. In this context, experiments were conducted using the CICIDS2017 dataset, a benchmark in the field of cybersecurity. Considering the tested algorithms and the CICIDS2017 dataset, none of the results obtained by the unsupervised techniques were satisfactory for the problem at hand. Therefore, it is concluded that more experiments and parameter adjustments are necessary to try to improve the performance of these algorithms and thus achieve reliable and accurate outlier detection.