Scaling Gaussian Process Regression with Full Derivative Observations

Daniel Huang

We present a scalable Gaussian Process (GP) method called DSoftKI that can fit and predict full derivative observations. It extends SoftKI, a method that approximates a kernel via…