Home | Trees | Indices | Help |
---|
|
The cross-entropy criterion is well suited for targets vector that are normalized between 0 and 1, used along with a final Sigmoid activation module.
It has been experimentally demonstrated that an AutoEncoder trains faster with a cross-entropy criterion.
The cross-entropy loss can be written as follows :
LCE = − 1 ⁄ N ∑Nk = 1(y⋅log(ŷ) + (1 − y)⋅log(1 − ŷ))
|
|||
|
|||
|
|
|||
Inherited from |
|
|
|
Home | Trees | Indices | Help |
---|
Generated by Epydoc 3.0.2 on Thu Aug 20 13:34:14 2015 | http://epydoc.sourceforge.net |