Add full covariance matrix support

[ebattenberg]

Will need to adapt sklearn full covariance routines to CUDAMat.

[ArtemiyFirsov]

Hey!

Is this repo going to be supported?

[ebattenberg]

Is this repo going to be supported?

Unfortunately, I won't be working on it, but I'd be glad to review contributions.

[ArtemiyFirsov]

@ebattenberg ok, thank you. And is are these scripts well tested?) If have some time, will work on that I guess)

[ArtemiyFirsov]

@ebattenberg BTW, when I try to estimate gmm from 100.000 samples with 26 dimensions of floats, I get CUDABlas error with no explanation. Using 1 GPU with 12GB of memory, it should be enough and ensured, that I shutdown cm sessions.

Actually, it works sometimes, not always))).

[ebattenberg]

I haven't run the code in about 4 years, so I can't guarantee that it will continue to work with current versions of CUDA. Sorry.

On Tue, Oct 10, 2017 at 11:24 PM Artemiy Firsov [email protected] wrote:

@ebattenberg https://github.com/ebattenberg BTW, when I try to estimate gmm from 100.000 samples with 26 dimensions, I get CUDABlas error with no explanation

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ebattenberg/ggmm/issues/2#issuecomment-335695535, or mute the thread https://github.com/notifications/unsubscribe-auth/ABUEeuDC9xuYWNhzc4yljCVcGuDwkcCaks5srF8agaJpZM4CRSEq .