Gibbs Parameter Sampling Algorithm Based on Finite Mixture Model of Random Point Pattern
Zhi Wang, Weifeng Liu, Yuxin Ding, Zilong Huang · 2020
A parameter sampling algorithm for finite mixture model of random point pattern by using the Gibbs sampling is proposed. First, the random finite set theory is used to construct finite mixture model of random point pattern. Secondly, using Gibbs sampling algorithm to estimate the parameters of mixture model. Finally, using BIC criterion to judge the effect of algorithm model fitting real data. In the simulations Gaussian mixture distribution is used as the characteristic distribution of random point pattern, several groups of random point pattern distribution are considered and compared with the traditional finite mixed model algorithm. The results show that the RPPFMM algorithm proposed in this paper is more accurate than the traditional FMM algorithm.