Initial feature set sampling using probabilistic Greedy algorithm

Sample initial solutions using a probabilistic version of Greedy algorithm.

Usage

sampleInitial(post_scores, constraints, size)

Arguments

post_scores

a vector of posterior scores (prior scores + likelihood) for each feature

constraints

a list containing feature-wise constraints

size

initial number of samples to be created. The output sample size can be lower, since duplicates are removed.

Value

a matrix containing initial feature sets as rows