Wednesday, April 4, 2018

Softmax vs Sigmoid function usage in last dense layer

If we require that sum of probabilities of all the classes should be equal to 1 then we use softmax but if we require separate and individual probabilities, such as in the case of multi-label classification we use sigmoid

https://www.depends-on-the-definition.com/classifying-genres-of-movies-by-looking-at-the-poster-a-neural-approach/



No comments:

Post a Comment