Title: A variational approach to a nonparametric density estimation using dependent species sampling models
Authors: Seongil Jo - Chonbuk National University (Korea, South) [presenting]
Abstract: The dependent species sampling model is a Bayesian nonparametric model for dependent data which estimates the probability density functions indexed by time or space. The dependent species sampling model is often applied a large data set, but the posterior sampling algorithm with Markov chain Monte Carlo can require lengthy computation time. We propose a variational method for the posterior approximation. We check the accuracy and speed of the variational method by simulation studies and illustrate the proposed method with some real data sets.