Finding Optimal Input Values for Desired Target Output by Using Particle Swarm Optimization Algorithm Within Probabilistic Models

Abstract

Developed predictive models, especially models based on probabilistic concept, regarding numerous potential combinatory states can be very complex. That complexity can cause uncertainty about which factors should have which values to achieve optimal value of output. An example of that problem is developed with a Bayesian network with numerous potential states and their interaction when we would like to find optimal value of nodes for achieving maximum probability on specific output node. This chapter shows a novel concept based on usage of the particle swarm optimization algorithm for finding optimal values within developed probabilistic models.