Artificial Neural Network (ANN) is my favorit algoritm because it use analogy of biological neural system. It is very interesting, neurons and axons convert to algorithm of transfer function and weight that can be adjust by activity called “learning”. When doing learning, ANN use some iteration to match the target with the input by changing it weight and some correction number, called bias.
Classification, Pattern Matching, signature identification, and other similar usage usually use ANN with some modification on input data and target. Because ANN has an ability to store memory on it neurons, some vendor use it for Pattern Matching application. There are some problem when using ANN, because when training, a lot of memory must be available if you don’t want your system being “hang”. The learning algorithm usually use the famous one, called backpropagation (from John Hopfield invention). But some research add other soft computing algorithms to enhanced the speed of learning and accuracy.
The language for creating ANN may be C++, Java or the easiest one, Matlab. I use Matlab for most of my application. Of course there is a classic problem by using Matlab, the lisence. But some language, for example, Octave, Scilab, etc, are open source and free for download. And the most important is the same syntax with Matlab. Of course we must build our function of ANN in open source language, because a lot of function (called M-file) does not available. If you want to know more about ANN with Matlab you can see www.mathworks.com or other millist, for example www.techsource.com.sg and free for sign up. This is a presentation form of explanation about ANN using Matlab you can download from techsource. Please find by your self in www.google.com.
Ok .. if you could not find, this is the link for download from my own storage: http://www.ziddu.com/download/16472066/neuralnetworkinmatlabtutorialpresentation.rar.html