search

isayev / ASE_ANI

ANI-1 neural net potential with python interface (ASE)
MIT License
220 stars 56 forks source link

Gaussian Activation Functions #9

Closed lilleswing closed 6 years ago

lilleswing commented 6 years ago

What mean and variance are you setting for your Gaussian Activation Functions?

Jussmith01 commented 6 years ago

The activation function as implemented in the model is:

f(x)=exp(-(z^2)) where z is the linear combination of the previous layes output with this nodes weights offset by a bias.