//
// PRIVATE METHODS
//
private BayesNet createWetGrassNetwork()
{
BayesNetNode cloudy = new BayesNetNode("Cloudy");
BayesNetNode sprinkler = new BayesNetNode("Sprinkler");
BayesNetNode rain = new BayesNetNode("Rain");
BayesNetNode wetGrass = new BayesNetNode("WetGrass");
sprinkler.influencedBy(cloudy);
rain.influencedBy(cloudy);
wetGrass.influencedBy(rain, sprinkler);
cloudy.setProbability(true, 0.5);
sprinkler.setProbability(true, 0.10);
sprinkler.setProbability(false, 0.50);
rain.setProbability(true, 0.8);
rain.setProbability(false, 0.2);
wetGrass.setProbability(true, true, 0.99);
wetGrass.setProbability(true, false, 0.90);
wetGrass.setProbability(false, true, 0.90);
wetGrass.setProbability(false, false, 0.00);
BayesNet net = new BayesNet(cloudy);
return(net);
}
public void testGibbsAsk_compare()
{
// create two nodes: parent and child with an arc from parent to child
IRandomVariable rvParent = new RandVar("Parent", new BooleanDomain());
IRandomVariable rvChild = new RandVar("Child", new BooleanDomain());
FullCPTNode nodeParent = new FullCPTNode(rvParent, new double[] { 0.7, 0.3 });
new FullCPTNode(rvChild, new double[] { 0.8, 0.2, 0.2, 0.8 }, nodeParent);
// create net
BayesNet net = new BayesNet(nodeParent);
// query parent probability
IRandomVariable[] rvX = new IRandomVariable[] { rvParent };
// ...given child evidence (true)
AssignmentProposition[] propE = new AssignmentProposition[] { new AssignmentProposition(rvChild, true) };
// sample with LikelihoodWeighting
ICategoricalDistribution samplesLW = new LikelihoodWeighting().Ask(rvX, propE, net, 1000);
Assert.AreEqual(0.9, samplesLW.getValue(true), DELTA_THRESHOLD);
// sample with RejectionSampling
ICategoricalDistribution samplesRS = new RejectionSampling().Ask(rvX, propE, net, 1000);
Assert.AreEqual(0.9, samplesRS.getValue(true), DELTA_THRESHOLD);
// sample with GibbsAsk
ICategoricalDistribution samplesGibbs = new GibbsAsk().Ask(rvX, propE, net, 1000);
Assert.AreEqual(0.9, samplesGibbs.getValue(true), DELTA_THRESHOLD);
}
public void testRejectionSample()
{
BayesNet net = createWetGrassNetwork();
MockRandomizer r = new MockRandomizer(new double[] { 0.1 });
Dictionary <String, bool> evidence = new Dictionary <String, bool>();
evidence.Add("Sprinkler", true);
double[] results = net.rejectionSample("Rain", evidence, 100, r);
Assert.AreEqual(1.0, results[0], 0.001);
Assert.AreEqual(0.0, results[1], 0.001);
}
public void testLikelihoodWeighting()
{
MockRandomizer r = new MockRandomizer(
new double[] { 0.5, 0.5, 0.5, 0.5 });
BayesNet net = createWetGrassNetwork();
Dictionary <String, bool> evidence = new Dictionary <String, bool>();
evidence.Add("Sprinkler", true);
double[] results = net.likelihoodWeighting("Rain", evidence, 1000, r);
Assert.AreEqual(1.0, results[0], 0.001);
Assert.AreEqual(0.0, results[1], 0.001);
}
public void testPriorSample()
{
BayesNet net = createWetGrassNetwork();
MockRandomizer r = new MockRandomizer(
new double[] { 0.5, 0.5, 0.5, 0.5 });
Dictionary <String, bool> table = net.getPriorSample(r);
Assert.AreEqual(4, table.Count);
Assert.AreEqual(true, table["Cloudy"]);
Assert.AreEqual(false, table["Sprinkler"]);
Assert.AreEqual(true, table["Rain"]);
Assert.AreEqual(true, table["WetGrass"]);
}
private static double enumerateAll(BayesNet net, List<string> unprocessedVariables,
Dictionary<String, bool> evidenceVariables)
{
if (unprocessedVariables.Count == 0)
{
return 1.0;
}
else
{
String Y = (String)unprocessedVariables[0];
if (evidenceVariables.ContainsKey(Y))
{
double probYGivenParents = net.probabilityOf(Y,
evidenceVariables[Y], evidenceVariables);
double secondTerm = enumerateAll(net, Util
.rest(unprocessedVariables), evidenceVariables);
return probYGivenParents * secondTerm;
}
else
{
double sigma = 0.0;
Dictionary<String, bool> clone1 = cloneEvidenceVariables(evidenceVariables);
clone1.Add(Y, true);
double probYTrueGivenParents = net.probabilityOf(Y,
true, clone1);
double secondTerm = enumerateAll(net, Util
.rest(unprocessedVariables), clone1);
double trueProbabilityY = probYTrueGivenParents * secondTerm;
Dictionary<String, bool> clone2 = cloneEvidenceVariables(evidenceVariables);
clone2.Add(Y, false);
double probYFalseGivenParents = net.probabilityOf(Y,
false, clone2);
secondTerm = enumerateAll(net, Util.rest(unprocessedVariables),
clone2);
double falseProbabilityY = probYFalseGivenParents * secondTerm;
// System.Console.Write(secondTerm + " ) )");
sigma = trueProbabilityY + falseProbabilityY;
return sigma;
}
}
}
public void testMCMCask2()
{
BayesNet net = createWetGrassNetwork();
MockRandomizer r = new MockRandomizer(
new double[] { 0.5, 0.5, 0.5, 0.5 });
Dictionary <String, bool> evidence = new Dictionary <String, bool>();
evidence.Add("Sprinkler", true);
double[] results = net.mcmcAsk("Rain", evidence, 1, r);
Assert.AreEqual(0.333, results[0], 0.001);
Assert.AreEqual(0.666, results[1], 0.001);
}
public static double[] ask(Query q, BayesNet net)
{
Dictionary<String, bool> evidenceVariables = q.getEvidenceVariables();
double[] probDist = new double[2];
// true probability
evidenceVariables[q.getQueryVariable()] = true;
probDist[0] = enumerateAll(net, net.getVariables(), evidenceVariables);
// false probability
evidenceVariables[q.getQueryVariable()] = false;
probDist[1] = enumerateAll(net, net.getVariables(), evidenceVariables);
// System.Console.WriteLine( probDist[0] + " " + probDist[1]);
// return probDist;
double[] normalized = Util.normalize(probDist);
// System.Console.WriteLine( normalized[0] + " " + normalized[1]);
return normalized;
}
/**
* Return a Dynamic Bayesian Network of the Umbrella World Network.
*
* @return a Dynamic Bayesian Network of the Umbrella World Network.
*/
public static DynamicBayesianNetwork getUmbrellaWorldNetwork()
{
FiniteNode prior_rain_tm1 = new FullCPTNode(ExampleRV.RAIN_tm1_RV,
new double[] { 0.5, 0.5 });
BayesNet priorNetwork = new BayesNet(prior_rain_tm1);
// Prior belief state
FiniteNode rain_tm1 = new FullCPTNode(ExampleRV.RAIN_tm1_RV,
new double[] { 0.5, 0.5 });
// Transition Model
FiniteNode rain_t = new FullCPTNode(ExampleRV.RAIN_t_RV, new double[]
{
// R_t-1 = true, R_t = true
0.7,
// R_t-1 = true, R_t = false
0.3,
// R_t-1 = false, R_t = true
0.3,
// R_t-1 = false, R_t = false
0.7
}, rain_tm1);
// Sensor Model
FiniteNode umbrealla_t = new FullCPTNode(ExampleRV.UMBREALLA_t_RV,
new double[]
{
// R_t = true, U_t = true
0.9,
// R_t = true, U_t = false
0.1,
// R_t = false, U_t = true
0.2,
// R_t = false, U_t = false
0.8
}, rain_t);
Map <RandomVariable, RandomVariable> X_0_to_X_1 = new HashMap <RandomVariable, RandomVariable>();
X_0_to_X_1.put(ExampleRV.RAIN_tm1_RV, ExampleRV.RAIN_t_RV);
Set <RandomVariable> E_1 = new HashSet <RandomVariable>();
E_1.add(ExampleRV.UMBREALLA_t_RV);
return(new DynamicBayesNet(priorNetwork, X_0_to_X_1, E_1, rain_tm1));
}
/**
* Return a Dynamic Bayesian Network of the Umbrella World Network.
*
* @return a Dynamic Bayesian Network of the Umbrella World Network.
*/
public static DynamicBayesianNetwork getUmbrellaWorldNetwork()
{
FiniteNode prior_rain_tm1 = new FullCPTNode(ExampleRV.RAIN_tm1_RV,
new double[] {0.5, 0.5});
BayesNet priorNetwork = new BayesNet(prior_rain_tm1);
// Prior belief state
FiniteNode rain_tm1 = new FullCPTNode(ExampleRV.RAIN_tm1_RV,
new double[] {0.5, 0.5});
// Transition Model
FiniteNode rain_t = new FullCPTNode(ExampleRV.RAIN_t_RV, new double[]
{
// R_t-1 = true, R_t = true
0.7,
// R_t-1 = true, R_t = false
0.3,
// R_t-1 = false, R_t = true
0.3,
// R_t-1 = false, R_t = false
0.7
}, rain_tm1);
// Sensor Model
FiniteNode umbrealla_t = new FullCPTNode(ExampleRV.UMBREALLA_t_RV,
new double[]
{
// R_t = true, U_t = true
0.9,
// R_t = true, U_t = false
0.1,
// R_t = false, U_t = true
0.2,
// R_t = false, U_t = false
0.8
}, rain_t);
Map<RandomVariable, RandomVariable> X_0_to_X_1 = new HashMap<RandomVariable, RandomVariable>();
X_0_to_X_1.put(ExampleRV.RAIN_tm1_RV, ExampleRV.RAIN_t_RV);
Set<RandomVariable> E_1 = new HashSet<RandomVariable>();
E_1.add(ExampleRV.UMBREALLA_t_RV);
return new DynamicBayesNet(priorNetwork, X_0_to_X_1, E_1, rain_tm1);
}
/**
* Return a Dynamic Bayesian Network of the Umbrella World Network.
*
* @return a Dynamic Bayesian Network of the Umbrella World Network.
*/
public static IDynamicBayesianNetwork getUmbrellaWorldNetwork()
{
IFiniteNode prior_rain_tm1 = new FullCPTNode(ExampleRV.RAIN_tm1_RV, new double[] { 0.5, 0.5 });
BayesNet priorNetwork = new BayesNet(prior_rain_tm1);
// Prior belief state
IFiniteNode rain_tm1 = new FullCPTNode(ExampleRV.RAIN_tm1_RV, new double[] { 0.5, 0.5 });
// Transition Model
IFiniteNode rain_t = new FullCPTNode(ExampleRV.RAIN_t_RV, new double[] {
// R_t-1 = true, R_t = true
0.7,
// R_t-1 = true, R_t = false
0.3,
// R_t-1 = false, R_t = true
0.3,
// R_t-1 = false, R_t = false
0.7
}, rain_tm1);
// Sensor Model
IFiniteNode umbrealla_t = new FullCPTNode(ExampleRV.UMBREALLA_t_RV,
new double[] {
// R_t = true, U_t = true
0.9,
// R_t = true, U_t = false
0.1,
// R_t = false, U_t = true
0.2,
// R_t = false, U_t = false
0.8
}, rain_t);
IMap <IRandomVariable, IRandomVariable> X_0_to_X_1 = CollectionFactory.CreateInsertionOrderedMap <IRandomVariable, IRandomVariable>();
X_0_to_X_1.Put(ExampleRV.RAIN_tm1_RV, ExampleRV.RAIN_t_RV);
ISet <IRandomVariable> E_1 = CollectionFactory.CreateSet <IRandomVariable>();
E_1.Add(ExampleRV.UMBREALLA_t_RV);
return(new DynamicBayesNet(priorNetwork, X_0_to_X_1, E_1, rain_tm1));
}
