[mlpack] Project about Low rank/sparse optimization using Frank-Wolfe

[Chenzhe Diao]

Hello everyone,

My name is Chenzhe. I am a 4th year Ph.D. student in Applied Mathematics
from University of Alberta in Canada. Part of my research is about image
recovery using over-complete systems (wavelet frames), which involves some
machine learning techniques, and uses sparse optimization techniques as one
of the key steps. So I am quite interested in the project about "Low
rank/sparse optimization using Frank-Wolfe".

I checked the mailing list from last year. It seems that there was one
student from GSOC16 interested in a similar project. Is that still not done
for some special difficulties? I took a brief look of the Martin Jaggi paper,
it seems that the algorithm is not complicated by itself. So I guess most
of the time for the project would be to implement the algorithm in desired
form, and to make extensive tests? What kinds of tests are we expecting?

Also, I checked src/mlpack/core/optimizers/ and I saw the GradientDescent
class implemented. I guess I need to write a new class in similar structure?

Thanks!

Best,
Chenzhe
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://knife.lugatgt.org/pipermail/mlpack/attachments/20170315/3cb604b1/attachment.html>