Replace in-house kernel implementations - Githubissues

edwinb-ai / LeastSquaresSVM

A Least Squares Support Vector Machine implementation in pure Julia

https://edwinb-ai.github.io/LeastSquaresSVM/dev/

MIT License

1 stars 1 forks source link

Replace in-house kernel implementations #24

Closed edwinb-ai closed 3 years ago

edwinb-ai commented 3 years ago

The current implementation only accepts an in-house (read, a locally developed implementation) of a RBF kernel.

This is sub-optimal, to say the least because there are better packages for this. See KernelFunctions.jl.

Furthermore, the more kernels we adopt, the larger the code base that needs to be maintained.

The proposal is simple:

[x] Remove in-house implementations.
[x] Make it so KernelFunctions.jl is a dependency.
[x] Refactor tests to verify this.
[ ] Create a new example where other kernels can be used.

edwinb-ai commented 3 years ago

Almost there, just need to create a new example showcasing the new kernels.