MEYER WAVELET NEURAL NETWORKS PROCEDURES TO INVESTIGATE THE NUMERICAL PERFORMANCES OF THE COMPUTER VIRUS SPREAD WITH KILL SIGNALS

This study shows the design of the Meyer wavelet neural networks (WNNs) to perform the numerical solutions of the spread of computer virus with kill signals, i.e. SEIR-KS system. The optimization of the SEIR-KS system is performed by the Meyer WNNs together with the optimization through the genetic algorithm (GA) and sequential quadratic (SQ) programming, i.e. Meyer WNNs-GASQ programming. A sigmoidal-based log-sigmoid function is implemented as an activation function, while 10 numbers of neurons work with 120 variables throughout this study. The correctness of the proposed Meyer WNNs-GASQP programming is observed through the comparison of the obtained and reference numerical solutions. For the consistency and reliability of the Meyer WNNs-GASQ programming, an analysis based on different statistical procedures is performed using 40 numbers of independent executions. Moreover, the use of different statistical operators like mean, median, minimum, standard deviation and semi-interquartile range further validates the correctness of the Meyer WNNs-GASQ programming for solving the SEIR-KS system.

Automated summary

This study designs Meyer wavelet neural networks (WNNs) for the numerical solutions of the spread of computer virus with kill signals (SEIR-KS system). The SEIR-KS system is optimized using Meyer WNNs in combination with genetic algorithm (GA) and sequential quadratic (SQ) programming, referred to as Meyer WNNs-GASQ programming. A sigmoidal-based log-sigmoid function is used as the activation function, with 10 neurons and 120 variables. The correctness and reliability of the proposed Meyer WNNs-GASQ programming is validated through comparison with reference numerical solutions and statistical analysis.