numpy - Fastest non negative matrix factorization (NMF) solver in python? -
i'm using sklearn's projectedgradientnmf , nimfa's lsnmf solvers factor sparse matrix. projectegradientnmf runs slower converges closer solution while lsnmf runs twice fast converges further solution (frobenius norm distance measure).
i'm curious current fastest or closest solvers available python community or there better option sparse matrix (the matrix sparse, not scipy.sparse)?
there benchmark here: https://github.com/scikit-learn/scikit-learn/pull/4852 pull request including coordinate descent solver mblondel https://gist.github.com/mblondel/09648344984565f9477a
what mean sparse not scipy.sparse? library from?
Comments
Post a Comment