MATLAB Central Newsreader - classification using ANN

classification using ANN

thajudin — Fri, 06 Jul 2007 02:09:19 -0400

Hi all,

I would like to classify two sets of data using Artificial Neural
network. I am just a beginner in this field.

I would like to get a sample code so that I can give 5 to 30 inputs
and two outputs. I would like like to have a flexible hidden layers
(from three hidden layers omwards).

I hope somebody can help me in coding this using Matlab.

thanking you in advance

thajudin

Re: classification using ANN

Srikanth — Thu, 05 Jul 2007 23:24:13 -0400

For one thing, you need just one output, since you have two sets of
data - a zero would represent one set, and a 1 would represent another
set.

Re: classification using ANN

thajudin — Fri, 06 Jul 2007 02:49:38 -0400

Srikanth wrote:
>
>
> For one thing, you need just one output, since you have two sets of
> data - a zero would represent one set, and a 1 would represent
> another
> set.
>
>

It is absolutely correct. Can I get a sample code for to start with

rgds

thajudin

Re: classification using ANN

Srikanth — Fri, 06 Jul 2007 00:05:34 -0400

Nope, I don't have any code as such... any NN book would give you some
algorithm (Greatest descent algo. is an easy one), or since you know
there are two groups, you could use a mean feature vector algorithm.
However, having more than 2 hidden layers makes the system extremely
complex, and I doubt it would be worth the effort in any system.

Re: classification using ANN

Greg Heath — Fri, 06 Jul 2007 16:39:52 -0400

On Jul 6, 3:05 am, Srikanth <s...@xdtech.com> wrote:
> Nope, I don't have any code as such... any NN book would give you some
> algorithm (Greatest descent algo. is an easy one), or since you know
> there are two groups, you could use a mean feature vector algorithm.
> However, having more than 2 hidden layers makes the system extremely
> complex, and I doubt it would be worth the effort in any system.

Having more than 1 hidden layer is seldom
necessary. See the comp.ai.neural-nets FAQ.

Hope this helps.

Greg