ANN resources on the Internet
A huge number of internet resources are devoted to artificial neural
networks. This page is just a small collection of starting points
-- you can find more through the usual search machines.
Sources of online information about ANNs suggested in this document:
-
Other lists
-
Tutorials
-
FAQs
-
Software tools
-
Source code
-
Repositories
-
ANN societies
1. Other lists of ANN-related resources
Many lists are much more complete and updated more often than this one.
Surfing around on them will give you a good idea of what's out there.
-
An extensive,annotated, maintained list is: NEuroNet
- Software at King's College, London, UK -- links to commercial and
free simulators, directories and newsletters.
-
List at l'Institut de technologie de l'information du Conseil national
de recherches du Canada: http://ai.iit.nrc.ca/subjects/Neural.html
-- pointers to ANN research organisations and other resources, with brief
descriptions
-
A list at the FUNET archive: http://ftp.funet.fi/pub/sci/neural/www/www-catalogue.html
-
A very extensive, annotated list
of available ANN software at Pacific Northwest National Lab, Richland,
Washington, USA.
- A very useful list of reinforcement learning links can be found
right here at the EPFL's LSL.
2. Tutorials
Some people have published online tutorials and course notes. For example:
3. FAQs
Several Usenet newsgroups feature discussions of various aspects of neural
networks. The main one is (if you have news access)
comp.ai.neural-nets.
Its FAQ is here).
There are many others. Find their FAQs by doing a search for "neural network"
at www.faqs.org:
4. Software tools for
developing NNs
Here are a few of the major software tools packages. Check NEuroNet, FUNET
etc. for more.
-
SNNS-
Stuttgart Neural Network Simulator
-
A complete simulator with graphical network editing and visualisation tools.
Nicely-documented, customizable, several platforms (Unix+X)
-
Netlab
Software
-
A library of Matlab® functions and scripts based on the approach
and techniques described in the book Neural Networks for Pattern
Recognition by Christopher M. Bishop (Oxford University Press, 1995)
-
MathWorks:
Neural Network Toolbox
A neural network development environment that requires MATLAB
5.2 or higher. Many architectures, learning and training rules. Visualization
tools, PCA pre-and post-processing, extensible and customizable. Can generate
C code. Commercial; expensive!
DELVE
- Data for Evaluating Learning in Valid Experiments
A standardised environment designed to evaluate the performance
of methods that learn relationships based primarily on empirical data.
Comprises
-
A software environment for dataset manipulation and statistical analysis
-
Several datasets
-
A repository of learning methods
5. ANN source code
If you're writing a neural network program, you might want to look at some
code other people have written. Here are just a few examples:
-
C
-
Neural
Networks at your Fingertips -- portable, self-contained ANSI
C programs for ADALINE, Backprop (time-series application), Hopfield, Bidirectional
associative memory, Self-organizing maps, etc.
-
The
Perceptron Applet -- uses Tk for the user interface
-
C++
-
Spiderweb
-- an object-oriented C++ library for designing, implementing, and using
artificial neural networks.
-
Java
-
Neural
Network Character Recognition Applet by Francesco Saverio Parlato
-- uses a backprop network to
recognize noisy 5x7 letters
-
OCHRE (Optical CHaracter REcognition) by Jason Tiscione -- another backprop network doing character recognition.
Nice tutorial and interface.
-
Web applets for interactive
tutorials on artificial neural learning by Fred Corbett -- The starting point for
several of our applets here (artificial neuron, perceptron learning, multilayer perceptron applets).
- Dynamic Associative
Neural Memory Simulator by David Clark -- performs many learning algorithms
-
Mathematica
Backpropagated
Neural Network -- nice package that provides functions to simulate
a backpropagated neural network, to classify exemplars based on network
results, and to examine graphically and numerically the dynamic and final
state characteristics of the network.
-
Simulating
Neural Networks with Mathematica, Electronic Supplement -- source code
for the programs in the book entitled "Simulating Neural Networks with
Mathematica" by James A. Freeman (Addison-Wesley, ISBN: 0-201-56629-X).
-
Matlab
Mathsource 5.0
Contributed neural network m-files
Mathsource 4.0
Contributed neural network m-files
6. Repositories
Some big software repositories have packages you can download.
7. Some ANN societies
Useful for locating publications and conference
information.
These societies publish the journal Neural
Networks:
The IEEE publishes several NN-related journals,
such as IEEE Trans. on Neural Networks, IEEE Trans. on Fuzzy Systems,
IEEE Trans. on Evolutionary Computation. Their web pages also
have links to other NN societies' home pages.
[Neural Java home page]
Last updated 21 April 1999 by Alix
Herrmann