Software by Kevin Murphy and students
- PMTK: probabilistic
written by Mark Schmidt, for optimization, structure learning,
inference in UGMs, and much more!
- UGM structure learning using group L1
regularization, supports MRFs, CRFs, etc.
- DAG structure learning using L1
regularization . Uses L1 to detect Markov blankets.
Can optionally perform local search over node orderings.
- Bayesian DAG learning,
Bayesian inference for directed acyclic graph structures
using MCMC and dynamic programming
- GMMVBEM: Variational Bayesian EM for Gaussian
array CGH analysis of single and multiple samples
matlab interface to WinBUGS
- Bayes Net Toolbox:
state and parameter estimation (inference and learning) for (directed) graphical models
- CRF toolbox:
inference and learning in conditional random fields
Contains code for loopy belief propagation and MRFs.
- HMM toolbox:
inference and learning in hidden Markov models
- Kalman filter toolbox:
inference and learning in linear dynamical systems
- MDP toolbox:
value and policy iteration for tabular Markov decision processes
- Graph visualization:
automatic layout of graphs (interface to GraphViz)
- KPMtools: miscellaneous functions,
needed by many of my toolboxes.
- KPMstats: statistics functions
for learning/sampling Gaussians/ multnomials, cluster weighted
- Graph theory toolbox:
simple graph algorithms like depth first
search, triangulation, etc.
- Fitting HMMs using Entropic priors,
based on Matt Brand's code.
the MNIST digit set
- blockmatrix class
a handy Matlab script to make web page timetables for classes.
- Datasets in Matlab format