Preprocessing for blacklisting Get the lowest pairwise scoring edges.

Get the lowest pairwise scoring edges represented as a blacklist matrix. This blacklisting procedure is motivated by Koller & Friedman (2003). This is rarely used now as we found that it blacklists edges that have significant dependencies but are not in the top \(n\) edges. We prefer the GetIncrementalScoringEdges method.

Usage

GetLowestPairwiseScoringEdges(scorer, n_retain)

Arguments

scorer

A scorer object.

n_retain

An integer representing the number of edges to retain.

Value

A boolean matrix of (parent, child) pairs for blacklisting.

References

1. Koller D, Friedman N. Being Bayesian about network structure. A Bayesian approach to structure discovery in Bayesian networks. Mach Learn. 2003;50(1):95–125.

Examples

data <- bnlearn::learning.test

scorer <- CreateScorer(
  scorer = BNLearnScorer, 
  data = data
  )
  
blacklist <- GetLowestPairwiseScoringEdges(scorer, 3)

blacklist_scorer <- CreateScorer(
  scorer = BNLearnScorer, 
  data = data,
  blacklist = blacklist,
  cache = TRUE
  )

# Randomly sample a starting DAG consistent with the blacklist. Then
# convert to a partition.
dag <- UniformlySampleDAG(colnames(data)) * !blacklist
partitioned_nodes <- DAGtoPartition(dag)

results <- SampleChains(10, partitioned_nodes, PartitionMCMC(), blacklist_scorer)