JuliaStats / NMF.jl

A Julia package for non-negative matrix factorization
Other
91 stars 34 forks source link

Initialization using randomized SVD #58

Closed ghost closed 3 years ago

ghost commented 3 years ago

Hi all,

SVD in src/initialization.jl can be replaced with the randomized SVD because NMF does not necessarily require the fully precise computation of SVD. In fact, scikit-learn adopts the method. The randomized SVD is faster than the original one, so this issue may be related to #43. Thanks.

Reference: https://scikit-learn.org/stable/modules/generated/sklearn.utils.extmath.randomized_svd.html