Research on Intrusion Detection Algorithm Based on Optimized CNN-LSTM

Abstract

With the development of computer network technology, network security has become more and more important, and intrusion detection has become an important means of network attack detection. In recent years, machine learning has played an irreplaceable role in many fields. In order to improve the accuracy of intrusion detection, many machine learning algorithms have been applied in intrusion detection models. Through the learning of training samples in KDDCUP99 intrusion data, this paper uses the relevant theory of neural network to construct an intrusion detection classification model based on optimized convolutional neural network and long short-term memory network, which is used to distinguish between normal state and various intrusion states. Convolutional neural networks, deep neural networks and traditional decision tree algorithms are compared in details in terms of accuracy and loss. The experimental results show that the prediction accuracy of the algorithm proposed in this paper is 0.972, and the test loss is 0.045, which effectively improves the classification accuracy of intrusion detection. Finally, the future development direction and prospects of the algorithm are prospected to further improve the security of computer networks.