2012/3/6 Mathieu Blondel <[email protected]>: > Even if they would be useful, I'd rather avoid projects like > "maintenance" or "speed things up". I think projects with a > well-identified goal are more likely to be accepted by the PSF. > > I like Olivier's proposals for SGD-based low-rank and non-negative > matrix factorization but I'd rather merge them into a single project. > For the non-negative constraint, I believe we just need to add a > clipping-to-zero step at each iteration. Vlad's idea of a Cython-based > non-negative least squares solver is very good and would be a nice > bonus if time permits. > > For students, I'd strongly recommend to submit a first pull-request to > scikit-learn *before* you apply for the GSOC. Last year, by the time > he applied for GSOC, Vlad had already merged his NMF code. This made > his application very credible and convincing.
We could indeed merge both proposals in a single GSoC proposal, but I would like to keep them as 2 separate steps with the two examples: group lens movie recommendation and out-of-core NMF for topic modeling on wikipedia text. We could also work on making the MiniBatchSparseDirectionaryLearning class work on sparse input (that might be a way to implement the out-of-core NMF with positive clipping, it don't know if it's a good idea or not from a theoretical point of view). Is someone else (@mblondel) interested in mentoring (some of) this? I think it would be good to have at least 3 different proposals with 3 different mentor names on the wiki page rather soon-ish to have scikit-learn accepted as a member project of the PSF organization for GSoC. -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel ------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/ _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
