MEASUREMENT OF DESIGNED DATASET USING LIGHT GBM FOR INTRUSION DETECTION SYSTEM

Abstract

The internet coins the term Network security as the major Network for various financial, trading, management, banking, and various others. The intrusion into this vast network seems to be a great challenge, and its detection is cumbersome for cybersecurity experts. Network Intrusion Detection is a technique that prevents unauthorized access to the network by analyzing the traffic in the field of network security. This paper compares the proposed Intrusion detection dataset RDKS19 with various other datasets like KDDCup 99, NSL-KDD, CICIDS2017, CIDDS001, and UNSW_15 for accuracy, precision, recall score, F1 score, false-positive rate, and computation time. Gradient boosting lightGBM algorithm used for classification of larger datasets due to its increased accuracy, fast training speed, and ability to manipulate missing values. Proposed dataset gives optimized results such as Accuracy (99.23), Precision (99.27), Recall (99.24), F1_Score (99.23), Computaion_time (4.153) and FPR (0.004).