Classification of Hybrid Intrusion Detection System Using Supervised Machine Learning with Hyper-Parameter Optimization

Main Article Content

Priya R. Maidamwar, Dr. Mahip M. Bartere, Dr. Prasad P. Lokulwar


Intrusion detection systems are the foundation of network security (IDS). In order to detect intrusions, IDSs keep a check on the system's activity and behaviour. Various IDS models, such as misuse detection and anomaly detection, can be used to identify attacks at all levels. For both known and undiscovered attacks, anomaly detection has a high rate of false positives, but misuse detection has a high rate of detection accuracy for only known assaults. This r presents an intrusion detection system that use machine learning to solve the shortcomings of current approaches.This research proposes the Hybrid IDS, which employs various supervised machine learning techniques in which Grid search hyper-parameter optimization for Binary and Multiclass classification systems with univariate feature selection is used. Random forest and the multi-layer perceptron neural network algorithm are utilised in supervised machine learning methods. UNSW-NB15, a dataset developed in 2015, is used to evaluate the suggested model's performance. Dataset splitting, data preprocessing, feature extraction and selection, and model training, hyper-parameter tuning, and classification are the four steps of the proposed hybrid intrusion detection algorithm. In terms of intrusion detection, the results obtained show that the suggested model is successful, and it is able to increase accuracy and minimise FAR. In addition, the time it takes to process a request is very minimal.

Article Details