Latent K-Tree Bayesian Networks Learner

lkbn is a tool for learning bounded tree-width Bayesian networks with latent variables

Installation and usage

Get, install and test:

	go get -u github.com/britojr/lkbn...
	go install github.com/britojr/lkbn...
	go test github.com/britojr/lkbn... -cover

Usage:

	lkbn --help
	Usage: lkbn <command> [options]

	Commands:

Examples:

... Learn parameters for a given structure:

cd examples/
lkbn ctparam -d example.csv -p parms.yaml -bi example#0.ctree -bo example#1.ctree

Sample clique tree structures:

ctsample -d example.csv -p parms.yaml

Parameters file:

YAML file containing specific parameters for the learning algorithms

Parameters file fields:

em_max_iters:		max number of EM iterations

Parameters file example:

./examples/parms.yaml
# structure sampling
treewidth     : 1
num_samples   : 3
latent_vars   : 7, 12

# parameters learning
em_max_iters  : 20
em_threshold  : 1e-5
em_use_parms  : true
em_init_iters : 2
em_restarts   : 4