Neural network implementation on C++

• Author: David Estévez-Fernández
• Released under GPL license , Dec 2012

0. Index

• 1.Introduction
• 2.Dependencies
• 3.Compilation
• 4.Usage
• 5.Version log

1. Introduction

Implementation of a neural network and a training mechanism using backpropagation algorithm and gradient descend for optimization.

To use that network a simple terminal interface has been implemented.

This is a work for a programming subject at University Carlos III of Spain.

2. Dependencies

There are no extra dependencies to make the neural network work, but in order to visualize the interface properly a terminal with ANSI escape secuences capabilities is required, such as xterm or the gnome default terminal, gnome-terminal.

3. Compilation

• Using QtCreator:

Install QtCreator : sudo apt-get install qtcreator

Run qtcreator and open the NeuralNetwork.pro file downloaded from this repository.

Press "build" button.

• Using g++:

In a terminal run: cd yourDownloadPath/NeuralNetwork

Then: g++ *.cpp -o NeuralNetwork

4. Basic Usage

• To run the code just change the working directory with cd and then type in a terminal:

./NeuralNetwork

• To select one option in the menu just type the index of that option (starting at 0) and press enter key.

• There are test input files at "../Sample data", named 'x1.txt', 'x2.txt', etc.

• For more information please read the documentation file.

5. Version log

• Version 1.0 Basic neural network with training mechanism and simple terminal interface.