AdaptiveGuard: Real-Time Cyber Threat Classification and Novel Attack Detection using AI and Machine Learning

With the rapid expansion and complexity of cyber threats, securing and ensuring reliability of modern communications networks pose an ever-increasing challenge. Traditional intrusion detection systems usually use specific rules that limit their effectiveness to detect unknown or new attacks. To fill this gap, in this paper, we investigate into a real-time network anomaly detection system using traditional machine learning algorithms. The system utilizes Logistic Regression, Random Forest, SVM, KNN, and Naïve Bayes classifiers to detect normal network behavior from the malicious one. Special focus is given to data preprocessing, feature identification and model tuning steps leading to the enhancement of both detection rate and reduction in false alarm. Experiments with the CICIDS2017 and NSLKDD benchmarks show that the proposed method outperforms previous approaches in detecting malicious activities including Distributed Denial of Service attacks, illegal access and malware. The results show that lightweight machine learning models could achieve accurate, scalable and practical solutions of real-time network security monitoring.