Combining additivity and active subspaces for high-dimensional Gaussian process modeling

Mickael Binois, Victor Picheny

Gaussian processes are a widely embraced technique for regression and classification due to their good prediction accuracy, analytical tractability and built-in capabilities for un…