News for hackers. Stuff that matters.

A comment on Generalization in Deep Learning [pdf].

Re:Generalization in Deep Learning [pdf] by briga on Tuesday October 17, 2017 @02:55AM (#15489149)
Why are they using MNIST to demonstrate generalization? Isn't it basically a toy problem at this point?
[ Parent | on HN ]
Re:Generalization in Deep Learning [pdf] by Iv on Tuesday October 17, 2017 @03:20AM (#15489221)
It is a problem that is well-known and easy to solve. Their point is to show that they offer a boost in terms of training speed.
[ Parent | on HN ]
Re:Generalization in Deep Learning [pdf] by benrbray on Tuesday October 17, 2017 @11:10AM (#15491563)
The empirical results aren't really the "point" of this paper.
[ Parent | on HN ]