09-03-2017, 11:00 AM
To effectively detect and classify network intrusion data, this work presents a general regression neural network (GRNN) based on the artificial immune algorithm with elitist strategies (AIAE). The AIAE-GRNN algorithm, with the aim of improving its adaptability and precision, is the elitist file and the elitist crossover with the artificial immune algorithm (AIA). In this work, mean square errors (MSE) were considered the affinity function. The AIAE was used to optimize the smoothness of the GRNN; Then, the optimal smooth factor was resolved and replaced in the GRNN. The intrusive data were classified. The document selected a GRNN that was optimized separately using a genetic algorithm (GA), particle swarm optimization (PSO) and fuzzy C-media clustering (FCM) to allow a comparison of these approaches. As shown in the results, the AIAE-GRNN achieves greater precision than the PSO-GRNN, but the AIAE-GRNN run time is long, which was first tested. FCM and GA-GRNN were eliminated due to their deficiencies in terms of precision and convergence.
To improve the speed of operation, the document adopted Principal Component Analysis (PCA) to reduce the dimensions of intrusive data. With the reduction of dimensionality, the PCA-AIAE-GRNN decreases in precision less and has a better convergence than the PCA-PSO-GRNN, and the PCA-AIAE-GRNN forward speed was relatively improved. The experimental results show that the AIAE-GRNN has a greater robustness and precision than the other algorithms considered and, therefore, can be used to classify the intrusive data.