Implementation of continuous functions to conditional probability description in probabilistic networks

This article presents an implementation of continuous functions for description of conditional probabilities in probabilistic networks. Utilization of continuous functions requires fewer computations during propagation phase and is a more natural way to express conditional probabilities than specification of matrices in the discrete approach. The present state of the knowledge restricts continuous functions only to those from the family of multivariate Gaussian distribution in which mean value of the variable is a linear combination of other variables. Author presents the algorithm which enables employment of any function to express mean value. The only condition is the ability to linearly approximate such function. Described implementation of the algorithm verifies its successful performance.