Abstract: Activation functions provide deep neural networks the non-linearity that is necessary to learn complex distributions. It is still inconclusive what is the optimal shape for the activation ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results