Security analysis of information in campus network based on improved back-propagation neural network

This study focuses mainly on the role of back-propagation neural network (BPNN) in the analysis of campus network information security. First, the BPNN algorithm was analyzed. Then, in order to improve the performance of the algorithm in security analysis, it was improved by the particle swarm optimization (PSO) algorithm by optimizing the initial weight of the BPNN. Taking some safety logs from the campus network of Zhuhai City Polytechnic in October 2019 as experimental data, the error and classification accuracy of the BPNN and improved BPNN were compared. The results showed that the error of the improved BPNN was obviously smaller than that of BPNN, and the false alarm rate and missing report rate were also lower, indicating that the algorithm had good reliability. From the perspective of accuracy, the improved BPNN had excellent classification in big data samples. It could achieve 92.57% accuracy under 25,000 data volume, while the accuracy of the original BPNN was only 78.36%. The experimental results prove that the improved BPNN has good classification performance and can accurately identify attack logs in a campus network, suggesting that the improved BPNN was effective in information security analysis and is conducive to improving the security of a campus network. It is worth further promotion and application. © 2021 Begell House Inc.. All rights reserved.