Research on Communication Network Intrusion Detection Algorithm Based on Deep Learning

This paper proposes a dynamic weight deep learning intrusion detection algorithm based on multimodal data fusion. By fusing multi-source data such as network traffic and logs, the attention mechanism is used to achieve dynamic weight adjustment, and intrusion detection is performed by combining the improved convolutional neural network with the long shortterm memory network. Experiments were conducted on the Knowledge Discovery and Data (KDD) Cup 99 and NSL-KDD datasets. The results show that the accuracy of the algorithm reaches 98.7 % and 97.5 %, respectively, which is 12.3 % and 10.8 % higher than the traditional support vector machine algorithm, and 8.1 % and 7.3 % higher than the ordinary convolutional neural network algorithm; the recall rate reaches 96.8 % and 95.2 %, respectively, and the F1 value is 97.7 % and 96.3 %, respectively. It shows good performance when dealing with various attack types such as denial of service attacks and port scanning. The dynamic weight adjustment and multimodal data fusion strategy effectively improve the algorithm's ability to identify complex network attacks, verifying the effectiveness and innovation of the algorithm.