Genetic optimization over Bayesian networks for diagnosis tasks

This is a Matlab program made to test the genetic algorithm proposed to find out the tests which should be made to get the highest certainty over the inference (diagnosis) made on a Bayesian network, but without letting the costs of such tests grow too much.
The program works through an easy to use GUI. Although it was designed to use a very simple example (the one shown in the article describing the algorithm and its uses), it is general enough to hold other examples as well.
In order to use this program, you will need to have the Bayes Net Toolbox for Matlab developed by Kevin Murphy installed in your computer.


 Screenshots

Click here to see a set of screenshots of the GUI


 Download

Get compressed m files for version 1.0
Click here to get the Bayes Net Toolbox for Matlab (needed by the program).

Last updated: april 13, 2004


 Documentation

There is a user manual available.
For documentation on the way the scripts and functions work, type "help x" with x the name of the script/function on Matlab, or check the source code (I think it is fairly commented).
To execute the GUI simply type "agbn" on Matlab.


 Author

Author: Rafael Peņaloza
Studied at ITAM in Mexico, and worked in KAMET research group. Currently, I am an independent researcher.
Current Contact: Rafael Peņaloza


 Level of Support

I will try to keep improving this program to make even more general. This will be done as time allows.
I will try to answer every question e-mailed to me. Bug fixes and new features will be done upon request and/or during improvements.


 License

This program is free to use and distribute for non-comercial use. I hope that it might help someone on their own research. I would also like to see this program grow better and capable of doing other tasks; for now, that will only be possible with the help of other people interested on it.



Hosted by www.Geocities.ws

1