Introduction to Artificial neural network (4)--aforge. NET Introduction

Sample program Download: Http://files.cnblogs.com/gpcuster/ANN3.rar
If you have questions, please refer to the FAQ
If you do not find a satisfactory answer, you can leave a message below:)

0 Catalogue
Introduction to Artificial neural network (1)--application of single-layer artificial neural network
Introduction to Artificial neural network (2)--the basic concept of artificial nerves
Introduction to Artificial neural network (3)--multi-layer artificial neural Network application example
Introduction to Artificial Neural networks (4)--aforge.net

1 Introduction
In this article, we will introduce a framework aforge implemented in C # that allows you to easily manipulate human networks, computer vision, machine learning, image processing, genetic algorithms, etc.

2 Neural Network Design Section Framework introduction
Here , I would like to emphasize: This code is very beautiful writing, a code such as the beauty of poetry, let me Charmed.

This piece of code is in the Aforge.neuro namespace.

This library file consists of 6 main parts:

• Neuron-an abstract base class for all neurons (neurons) that encapsulates some of the basic elements common to all Neuron: weights, output values, and input values. Other neuron are derived from this foundation.
• Layer-Represents a collection of neurons. This abstract base class encapsulates the commonality of layers.
• Network-Represents a neural net that is a collection of neuron ' s layers. This abstract base class provides the commonality of the network. The other network is derived from this foundation.
• Iactivationfunction-the interface of the activation function (activation functions). All the activation functions are derived from the interface
• Iunsupervisedlearning-Interface to the unsupervised learning algorithm-this type of learning system provides input only during the learning process and does not provide the desired output for that input. The goal of the system is to constantly find better solutions.
• Isupervisedlearning-Interface with tutor Learning (supervised learning) algorithm-This type of learning system provides input and expected output for that input during the learning process. The goal of the system is to continuously correct the network through actual output and expected output.

The relationships of these classes can be used to represent:

This file contains the following 2 neural network systems:

• Activation Network
• Distance Network

The following 5 learning algorithms are available to solve different problems:

• Perceptron Learning
• Delta Rule Learning
• Back propagation Learning
• SOM Learning
• Elastic Network Learning

3 code example
Sample program Download: Http://files.cnblogs.com/gpcuster/ANN3.rar

Instructions for use of the program and introduction to Artificial neural networks (1)-similar to a single-layer artificial neural Network application example, you can refer to http://www.cnblogs.com/gpcuster/archive/2008/05/22/1204456.html. However, the framework implementation in the aforge.net is used.

For an example of an and operation, the code is as follows:

Activationnetwork network = new Activationnetwork (new Thresholdfunction (), 2, 1);
Perceptronlearning teacher = new perceptronlearning (network);

double[][] input = new double[4][];
double[][] output = new double[4][];

Input[0] = new double[]{0, 0};
Output[0] = new double[]{0};
INPUT[1] = new double[]{0, 1};
OUTPUT[1] = new double[]{0};
INPUT[2] = new double[]{1, 0};
OUTPUT[2] = new double[]{0};
INPUT[3] = new double[] {1, 1};
OUTPUT[3] = new double[] {1};

Double error = 1.0;

while (Error > 0.001)
{
Error = Teacher. Runepoch (input, output);
}

4 Further studies

You can view the source code of the class library and refer to this article "Neural Network on C #"

5 Preview
We will use the framework presented in this chapter to solve a practical problem:)

6 Summary

In this chapter, we introduce an Ann library, through which you can easily build your own neural network, but you need to understand what type of network is needed to solve the problem, and what learning algorithm to use with it.

Introduction to Artificial neural network (4)--aforge. NET Introduction