Software Packages for Graphical Models / Bayesian Networks

Last updated 18 November 2002.
Click here for a French version of this page (not necessarily up-to-date).

Src = source code included? (N=no) If so, what language?
API = application program interface included? (N means the program cannot be integrated into your code, i.e., it must be run as a standalone executable.)
Exec = Executable runs on W = Windows (95/98/NT), U = Unix, M = Mac, or - = any machine with a compiler.
Cts = continuous-valued nodes supported? (D = discretized)
GUI = Graphical User Interface included?
Learns parameters?
Learns structure? CI = means uses conditional independency tests
Sample = sampling methods (e.g., likelihood weighting, MCMC) supported?
Utility = utility and decision nodes (i.e., influence diagrams) supported?
Free = Is a free version available? Yes, No or Restricted. (Commercial products often have free versions which are restricted in various ways, e.g., the model size is limited, or models cannot be saved, or there is no API; in this case, we say R for "restricted". Academic products are often only free for non-commerical use; in this case, we still call them free.)
Undir = are undirected or chain graphs supported (as well as directed graphs)?
Inference = which inference algorithm is used? jtree = junction tree, varelim = variable (bucket) elimination, MH = Metropols Hastings, G = Gibbs sampling, IS = importance sampling, sampling = some other Monte Carlo method, polytree = Pearl's algorithm restricted to a graph with no cycles, none = the program is only designed for structure learning from completely observed data

Name

Authors

Src

API

Exec

Cts

GUI

Params

Struct

Sample

Utility

Free

Undir

Inference

Comments

Lumina

N

Y

W,M

Y

Y

N

N

Y

Y

R

N

sampling

spread sheet compatible

U. Helsinki

C++

Y

U

Y

N

Y

N

Y

N

Y

N

MH

Generates C++ for MCMC.

U. Helsinki

Java

Y

WUM

Y

Y

Y

N

N

N

Y

N

?

Bayesian Naive Bayes classifier.

Nijman (U. Nijmegen)

N

N

W

N

Y

N

N

Y

N

R

N

?

-

Bayesia Ltd

N

N

-

D

Y

Y

Y

Y

N

R

N

jtree

Supervised and unsupervised learning, clustering, analysis toolbox, adaptive questionnaires, dynamic models

Bayesware Discoverer

Bayesware

N

N

WUM

D

Y

Y

Y

N

N

R

N

?

Uses "bound and collapse" for learning with missing data.

U. Helsinki

N

N

WUM

D

Y

Y

Y

N

N

Y

N

?

Runs on their server: view results using a web browser.

Motomura (ETL)

Java

Y

WUM

NN

Y

Y

N

N

N

Y

N

?

For learning, represents BN as a neural net.

Belief net power constructor

Cheng (U.Alberta)

N

W

W

N

Y

Y

CI

N

N

Y

N

?

-

Murphy (U.C.Berkeley)

Matlab/C

Y

WU

Y

N

Y

Y

Y

Y

Y

N

Many

Also handles dynamic models, like HMMs and Kalman filters.

BNJ

Hsu (Kansas)

Java

-

-

N

Y

N

Y

Y

N

Y

N

jtree and imptce sampling

-

Gowans (Imperial)

Visual Basic

Y

W

N

Y

N

Y

N

N

Y

N

Polytree

Parser and GUI for the XML-BIF format.

Rish (U.C.Irvine)

C++

Y

WU

N

N

N

N

N

N

Y

N

Varelim

-

MRC/Imperial College

N

N

WU

Y

W

Y

N

Y

N

Y

N

Gibbs

-

Business Navigator 5

Data Digest Corp

N

N

W

D

Y

Y

Y

N

N

R

N

Jtree

-

Cousins et al. (Wash. U.)

C

Y

WU

N

N

N

N

Y

N

Y

N

5 Sampling methods

-

Badsberg (U. Aalborg)

C/lisp

Y

U

N

Y

Y

CI

N

N

Y

Only

Jtree

Designed for contingency tables.

Poole et al. (UBC)

Java

N

WU

N

Y

N

N

N

N

Y

N

Varelim

-

Noetic Systems

N

N

WM

N

Y

N

N

N

N

R

N

?

-

ENS Lyon

Java

Y

WUM

N

N

N

N

N

N

Y

N

Jtree

-

Wilkinson (U. Newcastle)

C

Y

WUM

Only

N

N

N

Y

N

Y

N

Exact

Useful as a subroutine for Bayesian analysis of large linear Gaussian directed models.

Rue (U. Trondheim)

C

Y

WUM

Only

N

N

N

Y

N

Y

Only

MCMC

Bayesian analysis of large linear Gaussian undirected models.

U. Pittsburgh

N

WU

WU

N

W

N

N

Y

Y

Y

N

Jtree

-

Hugin

N

Y

W

Y

W

Y

CI

Y

Y

R

Y

Jtree

-

Warnes (U.Wash.)

Java

-

-

Y

Y

Y

N

Y

N

Y

Y

MCMC

-

Rockwell

Lisp

Y

WUM

N

Y

N

N

N

Y

Y

N

Jtree

GUI requires Allegro Lisp.

Cozman (CMU)

Java

Y

WUM

N

Y

N

N

N

Y

Y

N

Varelim/ jtree

-

HyperGraph Software

N

N

W

Y

Y

Y

Y

N

N

R

Y

Jtree

Up to 52 variables.

Microsoft

N

Y

W

N

W

N

N

N

Y

R

N

Jtree

-

Norsys

N

WUM

W

Y

W

Y

N

Y

Y

R

N

?

-

PMT

Pavlovic (BU)

Matlab/C

-

-

N

N

Y

N

Y

N

Y

N

special purpose

-

IRIDIA

Lisp

Y

WUM

N

Y

N

N

N

N

Y

N

?

Uses valuation systems for non-probabilistic calculi.

Dodier (U.Colorado)

Java

Y

WUM

Y

Y

N

N

N

N

Y

N

Polytree

Distributed implementation.

CMU

N

N

WU

Y

N

Y

CI

N

N

Y

Y

None

-

?

Java

-

-

N

Y

N

Y

N

N

Y

N

jtree

K2 for struct learning

Xiang (U.Regina)

Java

Y

WUM

N

Y

N

N

N

Y

Y

N

?

-

Microsoft

N

N

W

Y

Y

Y

Y

N

N

Y

Y

None

Learns BN or dependency net structure.

Cowell (City U.)

N

N

W

N

Y

Y

N

N

Y

Y

Y

Jtree

-

Other sites related to graphical models

Google's list of Bayes net software.
Decision analysis programs, list maintained by Marek Druzdzel.
Comparison of decision analysis software packages by Håkan L. Younes. The emphasis is on real-time decision making.
Older belief net programs, list created (but no longer maintained) by Russ Almond.
Repository of Bayes nets.
David Dowe's Mixture modelling page

Other sites related to Bayesian methods

Radford Neal's MCMC sofware (in C) for Bayesian neural networks, mixture models, Gaussian processes, etc.
Bayes linear methods
MCMC Preprint service
Bayesian Analysis, Computation, and Communication
Social network analysis software in R

See also