tag:blogger.com,1999:blog-6141980.post2933021334839171792..comments2024-03-20T12:28:35.004-05:00Comments on Nuit Blanche: Do Deep Nets Really Need to be Deep?Igorhttp://www.blogger.com/profile/17474880327699002140noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-6141980.post-75511855453091665272014-03-04T13:46:34.232-06:002014-03-04T13:46:34.232-06:00Leslie,
You are probably right, if the paper has ...Leslie,<br /><br />You are probably right, if the paper has not mentioned, it probably should have.<br /><br />Btw, I mentioned Yoshua Bengio work here: http://nuit-blanche.blogspot.fr/2013/12/sunday-morning-insight-sharp-phase.html<br /><br />Cheers,<br /><br />Igor.Igorhttps://www.blogger.com/profile/17474880327699002140noreply@blogger.comtag:blogger.com,1999:blog-6141980.post-50520412596784243112014-03-03T16:31:21.177-06:002014-03-03T16:31:21.177-06:00Sorry for the late comment to the post - your post...Sorry for the late comment to the post - your post yesterday reminded me of this question.<br />The paper "Do deep nets really need to be deep?" should have referenced the paper by Yoshua Bengio (one of the leaders in the field of deep learning) called "Learning Deep Architectures for AI" because the second section called "Theoretical limitations of shallow architectures" is very relevant to this question. It seems like there are strong reasons to use a deep architecture.Anonymoushttps://www.blogger.com/profile/06179125920265282125noreply@blogger.comtag:blogger.com,1999:blog-6141980.post-32709492399012823832014-01-04T00:34:33.302-06:002014-01-04T00:34:33.302-06:00Really Interestin, but it seems the SNN with the s...Really Interestin, but it seems the SNN with the same number of parameters performs worse than the DNN or CNN. To get comparable performance from the SNN they have to substantially increase the number of parameters.Anonymoushttps://www.blogger.com/profile/09661616788650836824noreply@blogger.com