Package edu.duke.cs.banjo.learner.components

Interface Summary
CycleCheckerI Documents the interface for creating a cycle checker implementation.
DeciderI Documents the interface for creating a decider implementation.
EquivalenceCheckerI Documents the interface for creating an equivalence checker implementation.
EvaluatorI Documents the interface for creating an evaluator implementation.
ProposerI Documents the interface for creating a proposer implementation.
RecorderI Documents the interface for creating a statistics implementation for describing the progress and results of the search algorithms.
 

Class Summary
CycleChecker Combines common code shared by the different cycle checker implementations.
CycleCheckerApplyThenCheck Checks for a cycle in the network by breadth first traversal.
CycleCheckerCheckThenApply Checks for a cycle in the network by depth first traversal.
Decider Combines common code shared by the different decider implementations.
DeciderGreedy Decides whether to keep a proposed change, based on a greedy decision algorithm.
DeciderMetropolis Decides whether to keep a proposed change, based on a Metropolis-Hastings decision algorithm.
EquivalenceChecker Combines common code shared by the different equivalence checker implementations.
EquivalenceCheckerBasic Provides basic equivalence checking between networks.
EquivalenceCheckerSkip Implements a "dummy" equivalence checker that lets us skip any equivalence comparisons during a search.
Evaluator Combines common code shared by the different evaluator implementations.
EvaluatorBDe Computes the BDe score of a network.
Proposer Combines common code shared by the different proposer implementations.
ProposerAllLocalMoves Proposes a list of all BayesNetChanges that can be applied in a single step (using addition, deletion, or reversal of a single edge), based on the current network configuration.
ProposerRandomLocalMove Proposes a single potential BayesNetChange based on the current network configuration.
Recorder Combines the common code shared by the different statistics implementations.
RecorderStandard Provides a set of "standard" statistics describing the progress and results of the search algorithms.