Intrusion detection model based on multi-kernel approximation and multi-layer neural network

Abstract

Aiming to address the problems of low time efficiency and poor generalization ability in support vector machine(SVM) model when dealing with large-scale network intrusions, this paper suggests a large-scale robust intrusion detection model that combines multi-layer neural network and multi-kernel approximate support vector machines. The multi-layer neural network carries out representation learning, learns the essential properties of the dataset, and performs feature reduction on the dataset. The multi-kernel approximation SVM using random Fourier characteristics to perform kernel approximation can handle large-scale datasets. The model employs the gradient descent method to train neural networks and multi-kernel SVM from start to finish. Our model was tested on three intrusion detection datasets of varying scales: NSL-KDD, UNSW-NB15, and CICIDS2017; and compared with conventional machine learning models, deep learning models, and SVM models of different variants. The experimental findings show that our model has higher classification performance and better robustness when processing large-scale datasets, and has more advantages in terms of time complexity.