Detecting APT Using Machine Learning: Comparative Performance Analysis With Proposed Model

This research explores the efficacy of machine learning algorithms in identifying Advanced Persistent Threats (APTs), pivotal in contemporary cybersecurity. Leveraging a dataset meticulously analyzed through Gradient Boosting, MLP-Classifier, Random Forest, and a Proposed Model, this study precisely scrutinizes each model's performance using metrics of accuracy, precision, recall, and Fl-score. The investigation showcases the superiority of the Proposed Model, demonstrating an accuracy of 97%, precision and recall of 0.97 and 0.98, respectively, alongside an Fl-score of 0.97. The Comparative Performance Analysis reveals nuanced strengths and limitations within each model, emphasizing the critical role of precision and recall in APT detection.