A Fast Direct Calculation Inference Algorithm

Abstract

AbstractAt present, among the inference algorithms on Dynamic Bayesian Networks (DBNs), the advantages of the Direct Calculation Inference (DCI) algorithm is that it needn’t perform the complex graphic transformation, its calculation formula is simple, and easy to program, however, the disadvantage is that the inference efficiency is quite low when there are many time slices. In this paper, after analyzing the algorithm complexity, we managed to find the crucial steps for decreasing the algorithm complexity, proposed a Fast Direct Calculation Inference (FDCI) algorithm based on optimizing the calculation means. It is proved by the simulation experiments that the inference results of the FDCI algorithm and the DCI algorithm are equal, but the inference efficiency of the proposed algorithm is much higher.Keywordsdynamic Bayesian networksinferencesoft evidencescomplexity