The relationship between Gaussian process based c-regression models and kernel c-regression models

Kernel regression and Gaussian process regression are known methods for representing non-linear regression models. The c-regression models is a method for obtaining the cluster partition and regression model simultaneously. The kernel c-regression models is a typical method of extending the c-regression models to the non-linear. This paper proposes a c-regression models based on Gaussian process regression as an approach to non-linearisation that differs from kernel c-regression models. Next, the relationship between the proposed method and the kernel c-regression models is presented. It is then shown experimentally that the proposed method and kernel c-regression models yield the same results under the same parameters and initial values.