A mixed-categorical correlation kernel for Gaussian process

Recently, there has been a growing interest for mixed-categorical meta-models based on Gaussian process (GP) surrogates. In this setting, several existing approaches use different strategies either by using continuous kernels (e.g., relaxation and Gower distance GP) or direct estimation of the correlation matrix. paper, we present kernel-based approach that extends exponential to handle variables. The proposed kernel leads new GP surrogate generalizes both models. We demonstrate, analytical engineering problems, our model gives higher likelihood smaller residual error than other state-of-the-art Our method is available in open-source software SMT.