100 citations · 104 across the 15 of their papers we have counts for
1 paper · 1 filter
Fernando Berzal
The loss function used to train a neural network is strongly connected to its output layer from a statistical point of view. This technical report analyzes common activation functi…