KernelFunctions.jl
KernelFunctions.jl copied to clipboard
Julia package for kernel functions for machine learning
Currently our build is broken due to a Zygote AD issue. To unblock other PRs, let's uncomment the test (see #414) and investigate the issue separately.
See #415
Here is a list of the failures in the tests made in #114 I observed with the different ADs : ForwardDiff.jl, Zygote.jl and ReverseDiff.jl : - [ ] FBMKernel :...
As it says :wink: The PR implements a modification of the suggestion in https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/issues/384 that is non-breaking and (hopefully) avoids most of the method ambiguity issues that I think could...
**Summary** We have a long-time problem for binary operations like `DotProduct` not satisfying the requirements of the `Distances.jl` framework (not a proper metric). Additionally, `Distances.jl` is very incompatible with GPU...
Adds a pair more kernels and benchmarks for AD Right now the issue is with ForwardDiff where I am not sure how to turn the parameters into a single `Array`,...
[Updated description] I reworked a bit more deeply the `spectral_mixture_kernel` and `spectral_mixture_kernel_product` functions: - `spectral_mixture_kernel` is now `SpectralMixtureKernel`. The previou implementation was false as it was expecting that `CosineKernel` was...
I wanted to ask for an overview of the current state of GPU support of this package. It appears as though there are several issues related to whether the package...
I don't know if this should be done in this PR. But `NystromFact` is awfully parametrized. _Originally posted by @theogf in https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/pull/427#discussion_r793744409_
We currently return a view to the underlying data when `getindex` is called on a `ColVecs`. See https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/blob/2d172125e72ca932133266c3b23d0865bec539df/src/utils.jl#L76 @theogf was the last person to touch this bit of code, but...