BNT Soft Discretization Package

By Imme Ebert-Uphoff
ebert  at  stups . com

The BNT Soft Discretization package is an
extension for Matlab's Bayes Net Toolbox BNT:
This Soft Discretization package provides the following functionality:
The training method consists of a soft discretization step that converts the continuous variables of the training cases into soft evidence, followed by a suitable parameter learning algorithm for the Bayesian Network.  

Most existing soft discretization approaches for Bayesian Networks use fuzzy set theory which is based on membership functions.  In contrast out method starts out with a probability density function that spreads the influence of a continuous variable to its neighbors, followed by a discretization step.  Thus our approach to soft discretization is based on probability theory, rather than fuzzy set theory.   It turns out that a membership function can be generated from the probability density function through convolution, yielding a set of probability-based membership functions.

Also available from the Georgia Tech smartech system:

Download of Version 1.0  (last updated Nov 10, 2009):

Select either option below to download the Matlab files:
Note: The source code download also contains the two pdf-files above.


I would love to get feedback from users.  Is this package useful?  Found a bug?
PLEASE drop me an e-mail with any comments you may have!

Return to main page:

Page last updated:  November 2009