News for hackers. Stuff that matters.

A comment on A Different Approach to Low-Rank Matrix Completion.

Re:A Different Approach to Low-Rank Matrix Completion by stared on Tuesday April 26, 2016 @03:01PM (#11574607)
Can anyone comment:

- if the results of the two methods are the same?

- how about the numerical stability? (many times tricky matrix algorithms have poor stability)

[ Parent | on HN ]
Re:A Different Approach to Low-Rank Matrix Completion by highd on Tuesday April 26, 2016 @03:13PM (#11574700)
Results should be the same as the convex optimization technique in the "well-defined" entries, though numerical stability is a real concern that I haven't looked at yet. Empirically I haven't had any issues, but if you have entries in the singular vectors sufficiently close to zero you can definitely imagine some problems. I don't think that's unique to this approach, either.
[ Parent | on HN ]