Systems Biophysics - Research - Self-Organizing Networks


	DemoGNG (Version 1.5)

Home

SysBio

Research

Networks

DemoGNG

The Institute
•	News
•	People
•	Teaching
•	Graduate study
•	Study and thesis projects
•	Jobs
•	Directions
•	Search
•	FTP Server
•	Contact
Chairs
•	Prof. Dr. Laurenz Wiskott
•	Prof. Dr. Gregor Schöner
Research groups
•	Theory of embodied cognition Prof. Dr. Gregor Schöner
•	Theory of Neural Systems Prof. Dr. Laurenz Wiskott
•	Neural Plasticity Lab PD Dr. Hubert Dinse
•	Real-Time Optical Imaging Lab Dr. Dirk Jancke
•	Organic Computing Dr. Rolf Würtz
•	Optimization of adaptive systems Jun.-Prof. Dr. Christian Igel
•	Autonomous robotics Dr. Ioannis Iossifidis
•	Medical Image Processing Dr. Susanne Winter
•	Real-time computer vision Jan Salmen
•	Multi-sensory fusion Dr. Andrey Bogdanov

(Please wait while loading ca. 122 KByte class-code.)

DemoGNG, a Java applet, implements several methods related to competitive learning. It is possible to experiment with the methods using various data distributions and observe the learning process. A common terminology is used to make it easy to compare one method to the other.

Your browser does not support Java applets.

Authors: Hartmut S. Loos Bernd Fritzke

Java Code

Model Description

Manual

Hopefully, the experimentation with the models will increase the intuitive understanding and make it easier to judge their particular strengths and weaknesses.

The following algorithms are available:

LBG (Linde, Buzo, Gray)
LBG-U (Fritzke)
Hard Competitive Learning (standard algorithm)
Neural Gas (Martinetz and Schulten)
Competitive Hebbian Learning (Martinetz and Schulten)
Neural Gas with Competitive Hebbian Learning (Martinetz and Schulten)
Growing Neural Gas (Fritzke)
Growing Neural Gas with Utility (GNG-U, Fritzke)
Self-Organizing Map (Kohonen)
Growing Grid (Fritzke)

If you are not quite familiar with what the applet does, have a look at the manual in HTML-format or as Postscript version (ca. 233 KByte).

For those of you, who want to go into further details: look at the technical report Some Competitive Learning Methods by Bernd Fritzke also available as Postscript version. Furthermore, you can Download the whole Package or just have a look at the DemoGNG Source or the Code Documentation.

DemoGNG (Version 1.5)