A PROBLEMÁTICA DO ATAQUE E O IMPACTO DO BALANCEAMENTO DE DADOS EM ALGORITMOS DE CLASSIFICAÇÃO DE ATAQUES

Abstract

The increasing prevalence of ransomware attacks has become a critical concern, largely driven by the growing dependence on the internet for communication and large-scale data exchange. Such attacks pose significant threats to the continuity of essential operations and the confidentiality of sensitive information. This study explores the application of machine learning algorithms for ransomware detection, with a particular focus on data balancing strategies aimed at improving the robustness and reliability of predictive models. An experimental methodology was adopted, employing a publicly available dataset to train and validate the models, which were assessed using both quantitative and qualitative performance metrics. The results indicate that data balancing significantly enhances detection performance, achieving accuracy rates exceeding 96%. These findings provide valuable insights for the advancement of cybersecurity solutions, particularly in the development of effective defense mechanisms against ransomware-based threats.