Practical Hilbert space approximate Bayesian Gaussian processes for probabilistic programming

[christineshen]

Hey! I’ve thought through the math and now I can see that out-of-sample prediction under HSGP is just like the linear model and there’s no need for matrix inversion.