Paper Title
Detecting DDos Attacks using Machine Learning Techniques

Abstract
Abstract - The large amount of connecting devices in critical zones such as militaries, banks, hospitals, and schools make detecting cyber anomalies an essential service. The interconnectivity of the computer network nowadays is highly subject to a security failure that stops the operation of the network. A Distributed Denial of Service attack (DDoS) is one of the most frequent threats to computer systems, as it disrupts services and causes network instability. One of the main challenges in detecting DDoS attacks is that it there many different types with different behaviours, confusing the monitor system on whether a service request is legitimate or not.. The proposed approach in this paper is to run a set of experiments to evaluate the two machine learning methods LightGBM and Random Forest on classifying DDos attacks. Relevant features are selected using XGBoost and Decision Tree in traditional and OvR Multi classification methods. To categorize and identify a DDoS attack, different types of attacks are used from the public dataset CICDDoS2019. Experiments show that using multiclassification OvR boosted the accuracy, and the best results were found in the classification using Random Forest with a selected feature of the Decision Tree and OvR multiclassification with an accuracy of 0.9942. Keywords - Machine learning, DDoS Attacks, Random Forest, LightGBM