This paper would be much improved by cutting the material on linear models, whic...

conjectures on May 25, 2016 | parent | context | favorite | on: Deep Learning Without Poor Local Minima

This paper would be much improved by cutting the material on linear models, which AFAIK is not new. Then by properly explaining the assumptions (Axa...) they are trying to do away with for nonlinear models rather than referring the reader to previous work.