neural network software

neural network freeware

Version 0.60 is out on October 10, 2008

This article is translated to Serbo-Croatian language by Web Geeks .

Last modified Thursday, 13-Sep-2012 07:21:33 CDT
[an error occurred while processing this directive]

These sites used this neural network freeware:
A Neural Network AMOS for SFO surface air temperature
Statistical prediction for SFO summer burnoff
You may want to visit them directly.

A simple demonstration

Here we demonstrate the effectiveness of the neural network with some "fake" data:

A successful neural network prediction of mock data in which the target in the data is a nonlinear function of the input vector, plus some noise. (Points on the diagonal are a perfect fit).

The singular value decomposition prediction, which shows the best possible linear prediction.

History of the mean square error. The fact that the error in the verification data is only a bit larger than the error in the training data indicates the network has been trained to the signal rather than the noise. The coefficient alpha used in the minimization scheme is also shown. (The change in the "weights" is proportional to alpha).

Free source code is available here for using the backpropagation algorithm to fit a neural network to pairs of input vectors and target vectors.

Download v0.60: nnet.tar.gz (1 mb, includes compiled binaries that may work for you!)

Requirements are a F90 compiler, and numerical recipes. The numerical recipes are needed only for the singular value decomposition prediction, not the neural network. But the numerical recipes software may cost you approximately $100. So grab your advisor's credit card, buy the software, download it, unzip it, untar it, and assign NRROOT in the Makefile to the recipes_f-90 directory. That is all there is to it.

A fortran compiler just might be free for you, if you can grab Intel Fortran Compiler. For plotting, you will need my simpleSVG. You will also need the usual Unix/Linux (Ubuntu 8.04 for me) software: inkscape, gnumake and python Here is an example of how to use the software:

gunzip nnet.tar.gz
tar -xvf nnet.tar
cd nnet
ls
cd scripts 
chmod u+x *.py
cd ../bin
make nn
make svd

now add nnet/exe and nnet/scripts to your path, preferably in your .bashrc

cd ../mock
chmod u+x dothis
dothis

Learning more:

Read all the "Readme" files, especially the one in the mock directory.
The python utilities have a little help function, e.g.:mseplot.py --help

A few notes:

hashplot.py can accurately make scatter plots containing overlapping points, by adjusting the area of the point. Read the comments within. Overlapping data happens quite frequently when making scatter plots with meteorological data (temperature vs. wind direction, for example). However, the "hash" capabilities are not needed with this mock data because the points don't overlap.
The neural network program is designed to work with data scaled between -1 and 1.

You also may want to investigate a neural network AMOS for SFO surface air temperature. To learn more about neural networks, see neural network FAQ. You can also grab my ancient seminar Neural Networks For Fun and Prophet. Comments? bfiedler@ou.edu


A successful neural network prediction of mock data in which the target in the data is a nonlinear function of the input vector, plus some noise. (Points on the diagonal are a perfect fit).	The singular value decomposition prediction, which shows the best possible linear prediction.