Skip to content

Prototype implementation of direct and factored deletion algorithms to learn Bayesian network parameters from incomplete data under the MCAR and MAR assumptions. These algorithms are consistent, yet they only require a single pass over the data, and no inference in the Bayesian network.

License

Notifications You must be signed in to change notification settings

UCLA-StarAI/Direct-Factored-Deletion

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Direct & Factored Deletion

Prototype implementation of direct and factored deletion algorithms to learn Bayesian network parameters from incomplete data under the MCAR and MAR assumptions. These algorithms are consistent, yet they only require a single pass over the data, and no inference in the Bayesian network.

Logo

Publication

Guy Van den Broeck, Karthika Mohan, Arthur Choi, Adnan Darwiche, Judea Pearl. Efficient Algorithms for Bayesian Network Parameter Learning from Incomplete Data, In Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence (UAI), 2015.

Contact

Website

http://reasoning.cs.ucla.edu/deletion/

Main contact###

Guy Van den Broeck
Department of Computer Science
UCLA
http://web.cs.ucla.edu/~guyvdb/
guyvdb@cs.ucla.edu

Contributors

Usage

The code takes as input Bayesian networks in the UAI file format. A simple example called fire_alarm.uai is shown below.

BAYES
6
2 2 2 2 2 2
6
1 0
1 1
2 0 2
3 0 1 3
2 3 4
2 4 5
2
 0.01 0.99
2
 0.02 0.98
4
 0.9 0.1 0.01 0.99
8
 0.5 0.5 0.99 0.01 0.85 0.15 0.0001 0.9999
4
 0.88 0.12 0.001 0.999
4
 0.75 0.25 0.01 0.99

License

This source code is licensed under the Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0

This software uses the inference library inflib.jar, which is provided by the Automated Reasoning Group at UCLA. inflib.jar is licensed only for non-commercial, research and educational use.

About

Prototype implementation of direct and factored deletion algorithms to learn Bayesian network parameters from incomplete data under the MCAR and MAR assumptions. These algorithms are consistent, yet they only require a single pass over the data, and no inference in the Bayesian network.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published