The Libra machine learning toolkit includes implementations of a variety of algorithms for learning and inference with Bayesian networks and arithmetic circuits:
Learning algorithms -- Structure learning for BNs and ACs; Chow-Liu algorithm; AC weight learning
Inference algorithms -- Mean field, belief propagation, Gibbs sampling, AC variable elimination, AC exact inference
Libra's strength is exploiting context-specific independence (such as decision tree CPDs) to allow exact inference in models with high treewidth.
- Changes to previous version:
Initial Announcement on mloss.org.
Other available revisons
Version Changelog Date 1.1.2d
Version 1.1.2d (12/29/2015):
- Minor fixes to scripts
- Published in JMLR ML-OSS!
February 4, 2016, 08:51:50 1.1.2c
Version 1.1.2c (6/24/2015):
- Libra can now be installed via OPAM as well. To install OPAM, see: http://opam.ocaml.org/doc/Install.html . Then run: "opam install libra-tk".
- Updated documentation to describe OPAM installation.
June 25, 2015, 00:10:25 1.1.2
Version 1.1.2 (6/10/2015):
- Switched to 2-clause BSD license; added license headers to all source files.
- Switched to OASIS build system for much cleaner compilation and installation procedures.
June 11, 2015, 08:07:21 1.1.1
Version 1.1.1 (5/21/2015):
- Many minor fixes to documentation, scripts, and code.
May 22, 2015, 10:16:52 1.1.0
Version 1.1.0 (3/26/2014):
- Added SPN library
- Added API documentation for all libraries
- Introduced libra script as a unified interface to all programs
- Many minor improvements to the code, interface, and documentation.
March 27, 2015, 06:29:45 1.0.1
Version 1.0.1 (3/30/2014):
- Several new algorithms -- acmn, learning ACs using MNs; idspn, SPN structure learning; mtlearn, learning mixtures of trees
- Several new support programs -- spquery, for exact inference in SPNs; spn2ac, for converting SPNs to ACs
- Renamed aclearnstruct to acbn
- Replaced aclearnstruct -noac with separate bnlearn program
- ...and many more small changes and fixes, throughout!
March 30, 2014, 09:42:00 0.4.0
Version 0.4.0 (7/06/2011): * MF inference in DNs (mf -depnet) * Max-product algorithm for BNs and MNs (maxprod) * MPE inference in ACs (acquery -mpe) * Added support for UAI MN file format. * New fstats utility to get basic file statistics for most file types supported by Libra * And more!
July 6, 2011, 09:40:25 0.3.0
Version 0.3.0 (8/01/2010):
- New data structure and functions for Markov networks with factors that are trees, tables, conjunctive features, or sets of conjunctive features.
- Added MN support to ACVE, BP, MF, Gibbs, and more.
- AC, BN, and MN scoring is now handled by a single program, mscore.
- Added mscore utility to convert between .xmod and .bif formats, or to go from .xmod/.bif to .mn (Markov network format).
- Added -noac option to aclearnstruct, so that it can be used to learn a Bayesian network that isn't represented as a circuit.
- Added dependency network learner (dnlearn)
- Extended tutorial, revised manual, and added more tests.
August 2, 2010, 07:21:28 0.2.0
Version 0.2.0 (6/08/2010):
- BP now supports table CPDs, not just trees
- Gibbs sampling now supports dependency networks with -depnet flag (experimental).
- Added -norb flag to disable Rao-Blackwellization in Gibbs sampling
- Fixed expat compilation under OS X
- Greatly expanded user manual
- Tweaks to the output of inference algorithms
- Added more automated tests, based on the tutorial
June 9, 2010, 00:43:28 0.1.0
Initial Announcement on mloss.org.
April 24, 2010, 11:38:24
No one has posted any comments yet. Perhaps you'd like to be the first?
Leave a comment
You must be logged in to post comments.