Abstract

Gaussian process regression (GPR) is a new machine learning method by the context of Bayesian theory and statistical learning theory. As a nonparametric method, GPR can adapt its complexity flexibly to the complex data distributions in the real-world, without any heuristics or manual tuning of parameters. Its hyper-parameters can be learned from data automatically. In addition, GPR can be computed efficiently because of its closed-form marginal probability computation. Since the inference of GPR is based on Bayesian rules, overfitting can be also avoided. In this paper, some historical tidal data are randomly selected as the training set to determine the hyper-parameters of the GPR model, and the others are used as the testing set to predict the tidal data. Finally, extensive experimental results show that GPR machine learning method can be used to predict the tidal data accurately and stably.

This content is only available via PDF.
You can access this article if you purchase or spend a download.