Robust t-Process Functional ANOVA Model-吴耀华 (中国科学技术大学)

Gaussian process regression model has already been used to fit data widely, and lots of its nice properties have been proved. In this talk, we combine a robust extended t-process with a Gaussian process in ANOVA functional model. Because of the extended t-process model’s clear construction, it inherits a lot of pretty properties from the GPR model. In this talk, robust property and information consistency of the Robust t-process functional ANOVA model are discussed. Simulation and real data studies present that the model has a robust fit about data and a good ability to predict it,compared with other functional methods.