Machine Learning based Automated System for Identifying Suspicious Network Usage

With the rapid increase in digital communication and the complexity of network infrastructures, detecting suspicious network usage has become a critical task. This study presents a machine learning-based automated system designed to improvise the identification of suspicious network activities. The proposed system works on supervised machine learning algorithms to investigate traffic in the network by identifying deviations from established patterns that may indicate potential security threats. As per our experimental results, CatBoost achieved the highest accuracy performance at 99.97% as well as exhibit perfect values across all metrics, with precision, recall, and F1 scores of 1.00 each. This research underscores the impact of learning models in identifying irregularities in the network which in future can provide a proactive defence mechanism against cyber threats.