Gaussian process mixture model for prediction based on maximum posterior distribution

Abstract

In this paper, a numerical experiment was proposed, which is based on the Bayesian procedure and Gaussian Process. Gaussian Process prior distribution was supposed as a before evaluating the hyperparameters of the Gaussian Process model via GyPytorch. Finally, the Gaussian process posterior was achieved and then confidence regions were visualized in two-dimensional planes. The Gaussian Process model could perform under common or specified kernel functions, while it depends on the prior information to a great extent.