System Attack Detection Using Machine Learning

In this research paper, anomaly detection in network traffic is addressed as a critical aspect of cybersecurity. Leveraging machine learning techniques on a benchmark dataset, we aim to develop a robust detection system. Our methodology involves analysing network traffic patterns and extracting features such as duration, protocol type, and service. Using labelled instances from the dataset, we train and validate the model, employing unsupervised learning to identify outliers and anomalies. Techniques like principal component analysis (PCA) enhance model performance and scalability. The system's effectiveness is evaluated using precision, recall, and F1-score, aiming to improve cybersecurity by accurately detecting network-based threats and minimizing false positives.