In the presentation below, local minima is regarded as not a problem in deep learning. Is this based on the practitioners' experience or is there any more formal justification?

http://bavm2013.splashthat.com/img/events/46439/assets/34a7.ranzato.pdf

asked Apr 18 '14 at 00:41

Ng0323's gravatar image

Ng0323
1567915

Be the first one to answer this question!
toggle preview

powered by OSQA

User submitted content is under Creative Commons: Attribution - Share Alike; Other things copyright (C) 2010, MetaOptimize LLC.