Abstract:
|
The log Gaussian Cox process is a flexible class of point processes for incorporating complex structures of space and time point patterns. As for Bayesian inference, for rich spatial point patterns, we need to sample high dimensional Gaussian latent variables and it would lead to heavy computational costs. Furthermore, hyperparameters for Gaussian latent variables have high correlation with Gaussian latent variables themselves, so standard MCMC sampling strategy is inefficient. In this paper, we propose an efficient and scalable computational strategy for spatial log Gaussian Cox processes. The key aspect of this approach is to calculate the approximate marginal posterior of parameters by utilizing pseudo-marginal approach.
|