NETWORK SECURITY ENHANCEMENT THROUGH MACHINE LEARNING BASED CYBER ATTACK DETECTION

Abstract

In today’s digital world, cyber-attacks such as DDOS, ransomware, and botnet attacks are increasing rapidly, creating a strong need for intelligent network security. An Intrusion Detection System (IDS) helps monitor network traffic and identify suspicious activities. Recent research shows that machine learning (ML) techniques can greatly improve intrusion detection because they learn patterns from data and accurately classify attacks. Different ML models—such as Naïve Bayes, SVM, KNN, Decision Tree, Random Forest, XGBoost, and Deep Learning—have been tested on modern datasets like UNSW-NB15 to compare their performance. Studies highlight that ensemble models and advanced algorithms like Random Forest and XGBoost often achieve high accuracy, precision, recall, and detection rates for modern cyber-attacks. Overall, ML-based IDS provides faster, more reliable detection of both known and unknown attacks, making it suitable for real-time network protection.