https://github.com/rory660/pynn

Python ANN Library
https://github.com/rory660/pynn

Last synced: about 2 months ago
JSON representation

Python ANN Library

• Host: GitHub
• URL: https://github.com/rory660/pynn
• Owner: rory660
• License: gpl-3.0
• Created: 2017-11-11T19:13:50.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2019-02-08T16:03:57.000Z (almost 6 years ago)
• Last Synced: 2023-10-27T12:35:07.197Z (about 1 year ago)
• Language: Python
• Homepage:
• Size: 28.3 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.md

Awesome Lists containing this project

README

        
# pyNN

Artifical Neural Network library for Python by Rory Brown  

This was built as a learning exercise.

# Documentation

### This module requires [Python 2.7](https://www.python.org/downloads/) and [NumPy](http://www.numpy.org/) to be used.  

## The NeuralNetwork Class

The NeuralNetwork Class allows for the creation of Neural Network objects, which are input, output and hidden layer configurable.  

### Constructor: NeuralNetwork(int inputSize, int outputSize, list hiddenLayerSizes)  

*inputSize* determines the number of input nodes for the network.  

*outputSize* determines the number of output nodes for the network.  

*hiddenLayerSizes* is a list containing the number of nodes for each hidden layer of the network.  

  

*NeuralNetwork(2,1,[3,3,3])* will generate a NeuralNetwork object with 2 input nodes, 1 output node, and 3 hidden layers with 3 nodes each.  

### forward(numpy.Array x), returns numpy.Array

This method takes an input *x* and uses the network to generate an output *y*.  

*x* is a NumPy Array object that contains input values for the network.  

Each row of the NumPy Array contains one set of input values.  

If the input layer is of size *n*, then each row much contain *n* values. Using multiple rows in the array will allow for multiple outputs to be calculated, which are returned, as *y*, in the same format as input *x*.