edu::tum::cs::srl::bayesnets::inference::SATIS::SampleSATPriors Class Reference

Inheritance diagram for edu::tum::cs::srl::bayesnets::inference::SATIS::SampleSATPriors:

Public Member Functions
	SampleSATPriors (PossibleWorld state, WorldVariables vars, Iterable<?extends AbstractVariable > db, BeliefNetworkEx bn) throws Exception
Protected Member Functions
void	setRandomState ()
Package Attributes
BeliefNetworkEx	bn
Random	generator
HashMap< BeliefNode, double[]>	priors = null

Detailed Description

SampleSAT samples uniformly from the set of solutions. If there is a large number of solutions and many of them are improbable, then the sampled distribution may be dominated by a few high-probability worlds that were sampled. To prevent this, we use the prior to initialize random variables, thus introducing a slight bias towards worlds with higher probability.

Author:: jain

Definition at line 93 of file SATIS.java.

Constructor & Destructor Documentation

edu::tum::cs::srl::bayesnets::inference::SATIS::SampleSATPriors::SampleSATPriors	(	PossibleWorld	state,
		WorldVariables	vars,
		Iterable<?extends AbstractVariable >	db,
		BeliefNetworkEx	bn
	)			throws Exception `[inline]`