Intelligent predictive stochastic computing for nonlinear differential delay computer virus model

Abstract

In this investigation, intelligent predictive stochastic computing is presented by exploitation of artificial neural networks Levenberg-Marquardt approach (ANNs-LMA) to analyze the dynamics of a nonlinear differential delay computer virus (DCV) model. The governing differential delay system with four classes representation with nonlinear delayed ordinary differential equations comprising of uninfected computers, latently infected computers, breaking out computers and computers having antivirus ability. The Adams approach for numerical solution is applied to produce the reference dataset by the variation in recruiting and detaching rate for both old as well as new PCs, bilinear transmission rate amongst healthy versus latently infected PCs, rate of latently infected PCs that break out, the rate for which breaking out PCs obtain antiviral ability, the rate for which antimalware PCs defeat all kind of viruses and delay with respect to time. The design ANNs-LMA is utilized to determine the approximate solution of the nonlinear DCV model by arbitrarily dividing the created dataset for training, testing as well as validation samples during learning of the networks. Negligible absolute errors, mean square errors, and relatively close to perfect modeling with regression metrics endorsed the strength, viability and reliability of the design ANNs-LMA for solving the nonlinear DCV models.