neural_network

neural_network allows you to create simple Neural Networks and to use genetic algorithm.

The code is open source, and available on github.

Installation

The simplest way to install neural_network is using pip:

pip install neural_network

Usage

First you need to include modules:

from neural_network import *

To create a new neural network, you need to instantiate the Net class.

net = Net(2, 3, 2)

net will be a neural network with 3 layers. 2 neurons on its input layer, 3 neurons on its hidden layer and 2 neurons on its outputs layer. You can create as many layer you want.

Documentation

neuralNetwork

class neural_network.Net(*topology)

Net class represent a neural network.

Parameters:

topology (*int) –

Integers representing the number of neurons on each layer

Exemple:

>>> net = Net(2, 1, 2)

feedForward(*inputs)

Calculate the outputs

Parameters:

inputs (*float) –

neural network’s imputs

Returns:

neural network’s outputs

Exemple:

>>> net = Net(2, 2)
>>> net.feedForward(0.5, 0.2)
[0.37993674654431087, -0.4970740393560804]

setInputsRange(minValue, maxValue, inputs=None)

Allow you to define the inputs range.

Parameters:

• minValue (float) – Minimum value for the inputs
• maxValue (float) – Maximum value for the inputs
• inputs (array) – Array of input’s index affected by setInputsRange

Exemple:

>>> net = Net(2, 3)
>>> net.setInputsRange(0, 100, [0]) # Changing input range just for the first input
>>> net.feedForward([0.5, 50])
[-1.0, 0.9999999999997584, 0.9941092385245458]

setOutputsRange(minValue, maxValue, outputs=None)

Allow you to define the inputs range.

Parameters:

• minValue (float) – Minimum value for the outputs
• maxValue (float) – Maximum value for the outputs
• outputs (array) – Array of ouput’s index affected by setOutputsRange

Exemple:

>>> net = Net(2, 3)
>>> net.setOutputsRange(0, 100) # Changing ouput range for all the outputs
>>> net.feedForward([0.5, 0.2])
[39.4994636910904, 50.68911915764991, 59.771121155018555]

geneticForNet

class neural_network.GeneticNet(nets)

Gentic algorithm class

Parameters:

nets (array) – Array of neural networks

Exemple:

>>> nets = []
>>> for i in range(10):
...     nets.append(Net([2, 3, 3]))
>>> alg = GeneticNet(nets)

setSelection(selectionStr)

Set the selection process

Parameters:

inputs (string) – name of the selection process (‘rank’ or ‘wheel’)

Exemple:

>>> net = Net(2, 2)
>>> net.setSelection('rank')