Bayesian networks based on Neural Networks
(Bayonet)
(Awarded by Japanese Society of Artificial Intelligence, 1998.12.9)
[On-line demonstrations of previous version(0.5)]
the previous version of BAYONET (ver0.5)
Older versions (Click to run the demo)
Example for a robot navigation (ver.0.2)
Example for a rainfall prediction
[Properties]
- Handling multi-valued discrete and continous variable data by neural networks.
- Learning from statistical data by neural networks learning algorithm.
- Generalization performance originate in interpolation of neural networks.
- Running as a server which can communicates to any other programs through TCP/IP.
- Concurrent calculation using multi-thread mechanisms of JAVA.
[Image of the simulator(ver.0.5)]
We have a plan to use this module for Office-Conversant Robot in ETL.
See
Jijo2 project,
(Japanese page).
[Download]
Newest version (ver 1.0) is stil developed now.
But, previous version has been opened to the public.
source codes (Applet version 0.5)
source codes (Applet version 0.2)
source codes (Applications(BNserver(TCP/IP) + GUII/F) version 0.2)
source codes (Application with Learning neural networks ver0.1)
This Applet is listed in the Java-Repository - a collection of documented
Java-ressources,
.
Other information
Other links about Bayesian Networks
Abstract of Bayesian Net(in Japanese)
Author's page
MOTOMURA Yoichi