Questions of convergence and computational time
Opened this issue · 0 comments
xmlyqing00 commented
Thank you for your attention. I have two questions.
(1) Why the fit process converges so quickly? It converges in the first two or three iterations. Is the reason that the initialize method makes a good initialization?
(2) I've compared this package with sklearn.decomposition.DictionaryLearning. This package does well in computational time which is much quicker than DictionaryLearning. I've noticed that this package implements a quicker version of KSVD paper. Is this the reason?