VIBES: Variational Inference for Bayesian Networks

What is VIBES?

VIBES is a software package which allows variational inference to be performed automatically on a Bayesian network (if the terms in italics don't mean anything to you, read this tutorial before continuing). I created VIBES during my Ph.D. as an implementation of my Variational Message Passing algorithm. For technical details of this algorithm, please see the relevant publications on my web site.

News:	18th Feb 2004	VIBES 2.0 available for download
	14th Oct 2004	VIBES 2.0.1 - allows using VIBES directly from within Matlab (beta version)

Getting Started with VIBES

Getting started with VIBES is a three-step process:

1.	Install Java Web Start:
	Java Web Start allows you to run applications directly from within your browser.
2.	Run VIBES 2.0.1.
	On first download, you will need to agree to VIBES having unrestricted access to your machine. In future, you can run VIBES directly from your local machine (using Web Start).
3.	Follow the VIBES Tutorial.
	The tutorial provides an introduction to using VIBES and includes example Bayesian Networks.

VIBES is released under the revised BSD license.

Running VIBES without Web Start

If Web Start doesn't work for your platform, then you can simply download the JAR file for the application and run it using:

java -jar Vibes2_0.jar

You must have Java installed on your system first - if you don't, go to java.com.