Daniel Dodd comments

Results 18 comments of


                                            Daniel Dodd

dev: VI: Refactor natural gradients, add dual parameterisation, add pseudo data.

As an aside to the above, it would be nice to add a high-level sklearn function to train and quickly "update" (or fantasise) new data for use in active learning....

feat: vector-valued GPs

Hi @thomascerbelaud, this has been made public, [JaxGaussianProcesses/MOGPJax](https://github.com/JaxGaussianProcesses/MOGPJax). Note this currently only has a `GPLVM` model. The `JaxLinOp` integration into GPJax has broken the multi-output prior and posterior stuff, particularly...

dev: Transformations.

Thanks for spotting this @murphyk (and apologies for the slow reply)! We recently refactored `GPJax`, first by removing transformations from all objects in `v0.5.0`. Then, our most recent release `v0.5.2`...

feat: Adding several acquisition functions

Thanks @miguelgondu - this would make an excellent contribution!

Add changepoint kernel

Nice work @jakeyeung! Just to check: have you rebased from the main branch - the change log seems huge here? Would just make it easier to see what you have...

Add changepoint kernel

Thanks @jakeyeung. Overall this looks really good. Quite excited to give this a go on some data soon. Mainly asking some clarification questions in the above review. Just the only...

bug: Jitting objectives.

Update: going for the design of option (b)(i), i.e., ```python # e.g., def elbo(negative: bool ...) -> callable: def elbo_fun(model, batch) -> Float[Array, ""]: ... return elbo_fun ```

dev: Use kwargs only in dataclasses

Tagging #372 as closely related issue.