Welcome to

Data On Stage.com

This site is dedicated to providing resources for Bayesian Networks and related techniques.

What are Bayesian Networks?
Bayesian Networks (BNs) are a tool for modeling systems containing uncertainty that has gained much popularity in recent years.  BNs use tools from probability theory (primarily Bayes' theorem, which gave them their name) to solve various tasks in the areas of data mining and artificial intelligence.

1) Soft Discretization for Bayesian Networks

The most recent contribution of this site is a research report on how to implement soft discretization for Bayesian Networks and the implementation of the corresponding algorithms as the BNT Soft Discretization Package.

A) Research Report Title:  A Probability-Based Approach to Soft Discretization for Bayesian Networks

Description:
This report discusses
• How to implement soft discretization to train a discrete Bayesian Network directly from continuous data.
• How to use soft discretization also for inference and how to convert the inference results from the discrete network to meaningful continuous output values.
The 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.

Last updated:  Sept 22, 2009

This file can also be downloaded from the Georgia Tech smartech system:  http://smartech.gatech.edu/handle/1853/30197

B) Implementation available: BNT Soft Discretization Package - NEW!

Last updated: Nov 10, 2009

Another contribution is a visualization tool for link strength and connection strength in discrete Bayesian Networks.

Description:
The LinkStrength / LinkConnectionStrength package calculates and visualizes link strengths and connection strengths in discrete Bayesian Networks.  Measures include Entropy, Mutual Information, True Average Link Strength and Blind Average Link Strength.

Implementations of this package are available for
1. BNT - Matlab's Bayes Net Toolbox
1. PNL - Intel's Open-Source Probabilistic Networks Library:
Available PNL packages:  PNLtoGraphviz Interface and LinkConnectionStrength Package

Research Reports available:

Last updated:  Sept 22, 2009

3) MarkovEquivalent Package

Description:
The MarkovEquivalent package contains a few functions to generate
• all DAGs represented by a pattern (partially oriented DAG) or
• all DAGs that are Markov equivalent to a DAG.
Implementation of this package is available only for BNT (Matlab's Bayes Net Toolbox) at MarkovEquivalent Package.

Last updated:  Feb 3, 2007

This site and its content are maintained by Imme Ebert-Uphoff
NEW e-mail address:  iebert at engr . colostate . edu!