CODEWITHSUNDEEP
algorithmartificial-intelligenceneural-network
shoen
shoen

Reputation: 11865

Artifical Neuron's Output

In the iteration1 image of the following wiki you can see a basic neural network: http://www.heatonresearch.com/wiki/Back_Propagation

If you look at the hidden layer's first neuron (H1) you'll notice that the sum is: -0.5313402159445314 and the output is: 0.3702043582229371

What I can't figure out is how the output was calculated. If I use sigmoid or hyperbolic tangent function on the sum value i get different results.

Thanks

Upvotes: 0

Views: 71

Answers (1)

Peter de Rivaz
Peter de Rivaz

Reputation: 33509

They are using the Sigmoid activation function.

The formula is 1/(1+exp(-x)).

We can check in Python via:

from math import exp
x=-0.5313402159445314
print 1./(1.+exp(-x))

Prints 0.370204358223

Upvotes: 2

Related Questions