This is a Standard RNN library.

Activation Function

Possible activation functions:

• Linear: identity, except that it truncates values to lie in [-1, 1]
• Sine: calculates the sine-wave f(x) = sin(x)
• Gaussian: calculates a gaussian f(x) = e^(-x*x), scaled to lie in [-1, 1]
• Sigmoid: calculates the sigmoid f(x)=tanh(x*lambda)

Testing the Library

g++ main.cpp -o RNN

./RNN [number_of_inputs] [number_of_outputs]

Visualizing the Neural Network

• Initial RNN

python plotRNN.py std_RNN_ini.csv

• Evolved RNN

python plotRNN.py std_RNN_mut.csv