Comparative Study Between Adaptative Approaches for Intrusion Detection

“Today, internet security has become a matter of utmost importance. Intrusion detection represents a novel strategy for safeguarding data networks that were previously utilized. Deep learning and machine learning have gained widespread acceptance in enhancing the efficiency of intrusion detection systems. Even though these systems are continually evolving, training static classifiers in batches can identify intrusions without regard to the evolving characteristics of regular data streams. Moreover, in real-world intrusion detection systems, the data distribution is often transient, which gives rise to the phenomenon known as ‘concept drift’ or ‘transient learning.’ Concept drift refers to how people alter their opinions in online supervised learning environments and how changes in input data and target variables are connected over time. In this study, we conduct an in-depth examination and comparison of an Adaptive Random Forest (ARF) classification model equipped with ADWIN and Support Vector Regression (SVR) in contrast to previous approaches.”