Extending heteroskedastic likelihood example with kernel with different active dimensions #1892
Unanswered
dataforager
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello,
First time posting. Thanks for the great package.
I'm interested in extending the heteroskedastic likelihood (HL) notebook to model the location of my regression output with a latent GP in which the kernel is a sum of kernels using different active dimensions/subsets of the input space.
In the current HL notebook, SVGP is used. But wouldn't that mean that the approximation to the latent GP is expecting index points of the full input space dimensionality? leading to ValueErrors for shape mismatches?
Basically, is the model formulation I'm proposing possible?
likelihood = gpf.likelihoods.HeteroskedasticTFPConditional( distribution_class = tfp.distributions.Normal, scale_transform=tfp.bijectors.Exp(), )
Thanks for your help.
Beta Was this translation helpful? Give feedback.
All reactions