All sounds good. Looking forward to more concrete versions of this: class diagrams, proof of concept PRs, etc :)
Am Mo., 23. März 2020 um 15:53 Uhr schrieb sai_ng <jonpsy...@gmail.com>: > I'll do that under GSOC :) . I'm including that in my proposal, In my > proposal I'll be including the previous class diagram and my proposed class > diagram. I'll be glad if you see it Heiko ! :D. > Basically, what I'd rather go by is: > We could rather have separate classes for RFF,RBF(Random > binning..),KRR,Nystrom and then they could inherit from a super class, say > "CKernelApproximator". And our CKernelApproximator could inherit from > CKernel, how does that sound? > Doing this gives the advantage of having more freedom to add methods > suitable for different types of situations, like Primal and Dual > formulations. This would also make it more structured and in general more > Object Oriented. > > Thoughts? > > On Mon, Mar 23, 2020 at 9:08 PM Heiko Strathmann < > heiko.strathm...@gmail.com> wrote: > >> Yes that is a good point. It would be cool to have it implemented say for >> all subclasses of KernelMachine. All it really does is changing the basis >> set to represent the kernel function using landmark points rather than the >> training data. If you want to impement it in this manner, this would be a >> very welcome contribution. However, doing this in general is difficult, and >> e.g. SVMs will have a different implementation as the solver itself will be >> changed, so definitely checking out sklearn would help here for >> abstractions. >> Any ideas how to go ahead with this? >> H >> >> >> Am Mo., 23. März 2020 um 15:06 Uhr schrieb sai_ng via shogun-list < >> shogun-list@shogun-toolbox.org>: >> >>> Hi again, >>> It's me Nanubala Gnana Sai. I was checking out SKlearn implementation of >>> Nystrom and tried comparing with our own implementation. I was wondering, >>> why is the given approximation technique bounded to a specific method ( >>> for example: KRR), it should be implemented as different class atleast >>> that's what's done in Sklearn. There is an implementation for RFF which >>> works in a similar fashion, so I thought it's only logical to have Nystrom >>> implementation as well. Looking forward to hearing from you ! :D >>> >>
