Improved PSO-FNN for Network Security Node Optimization

Abstract

The network operation data contains a large amount of non-digital information; the traditional neural network cannot be used directly to obtain the network security situation. Reverse propagation due to its inherent shortcomings, neural network structure is often slow training speed, low learning efficiency, prediction accuracy is not high, so it does not meet the fuzzy layer centre of a fuzzy neural network to realize the clustering of input samples and determine the membership centre of the fuzzy neural network, and finally optimize the fuzzy neural network, get the PSO-FNN model, and the model is applied to the acquisition of network security elements. When the particle number starts at N=5, the number of particles, detection rate begins to increase gradually, When N=30, The detection rate reaches 83.856%, Time-consuming is about 308.13s. The rapid evolution of network technology has introduced unprecedented challenges in ensuring network security. As cyber threats become more dynamic and sophisticated, the need for robust and efficient solutions to optimize security nodes within networks is critical. In this context, this study proposes an improved model that integrates Particle Swarm Optimization (PSO) with Fuzzy Neural Networks (FNN), termed as PSO-FNN. This model aims to enhance detection accuracy and computational efficiency by leveraging the optimization capabilities of PSO for feature selection and parameter tuning, thereby addressing the limitations of traditional network security optimization techniques.