A MACHINE LEARNING-BASED HYBRID MODEL FOR BOTNET DETECTION

Abstract

Abstract: New hybrid approaches using machine learning algorithms are being developed and tested in response to the growing dangers posed by bots and botnets. This study examines two state-of-the-art models that include hybrid techniques: Enhanced Random Forest (ERF) and Enhanced Support Vector Machine (ESVM). Careful planning went into the study's execution in order to comprehend and contrast the two models on three separate datasets: the UCI Botnet Dataset, the Kaggle-NSLKDD Botnet Dataset, and the Data World Botnet Dataset. Part one of the research delves further into ERF. Designed to manage the intricacies of bot identification, this model is an improvement over conventional random forests. Furthermore, a strong comprehension of how these models behave under different circumstances is added by comparing them across the chosen datasets. In this comprehensive comparison, we see how ERF and ESVM are two of the most important advanced tools for detecting bots and botnets. Thus, the research lays the groundwork for further development of machine learning methods to detect and counteract bot attacks.