@Override
public String getParameterDescription(SufficientStatistics parameters, int numFeatures) {
Tensor featureWeights = getFeatureWeights(parameters);
TableFactor featureValues =
new TableFactor(
initialWeights.getVars(),
initialWeights.getWeights().replaceValues(featureWeights.getValues()));
List<Assignment> biggestAssignments =
featureValues.product(featureValues).getMostLikelyAssignments(numFeatures);
return featureValues.describeAssignments(biggestAssignments);
}
@Override
public TableFactor getModelFromParameters(SufficientStatistics parameters) {
Tensor featureWeights = getFeatureWeights(parameters);
double[] logProbs = featureWeights.getValues();
double[] probs = new double[logProbs.length];
for (int i = 0; i < logProbs.length; i++) {
probs[i] = Math.exp(logProbs[i]);
}
return new TableFactor(
initialWeights.getVars(), initialWeights.getWeights().replaceValues(probs));
}
@Override
public void incrementSufficientStatisticsFromAssignment(
SufficientStatistics gradient,
SufficientStatistics currentParameters,
Assignment assignment,
double count) {
Preconditions.checkArgument(assignment.containsAll(getVars().getVariableNumsArray()));
Assignment subAssignment = assignment.intersection(getVars().getVariableNumsArray());
long keyNum =
initialWeights
.getWeights()
.dimKeyToKeyNum(initialWeights.getVars().assignmentToIntArray(subAssignment));
int index = initialWeights.getWeights().keyNumToIndex(keyNum);
((TensorSufficientStatistics) gradient).incrementFeatureByIndex(count, index);
}
@Override
public void incrementSufficientStatisticsFromMarginal(
SufficientStatistics gradient,
SufficientStatistics currentParameters,
Factor marginal,
Assignment conditionalAssignment,
double count,
double partitionFunction) {
if (conditionalAssignment.containsAll(getVars().getVariableNumsArray())) {
// Short-circuit the slow computation below if possible.
double multiplier = marginal.getTotalUnnormalizedProbability() * count / partitionFunction;
incrementSufficientStatisticsFromAssignment(
gradient, currentParameters, conditionalAssignment, multiplier);
} else {
VariableNumMap conditionedVars =
initialWeights.getVars().intersection(conditionalAssignment.getVariableNumsArray());
TableFactor productFactor =
(TableFactor)
initialWeights
.product(
TableFactor.pointDistribution(
conditionedVars, conditionalAssignment.intersection(conditionedVars)))
.product(marginal);
Tensor productFactorWeights = productFactor.getWeights();
double[] productFactorValues = productFactorWeights.getValues();
int tensorSize = productFactorWeights.size();
double multiplier = count / partitionFunction;
TensorSufficientStatistics tensorGradient = (TensorSufficientStatistics) gradient;
for (int i = 0; i < tensorSize; i++) {
int builderIndex = (int) productFactorWeights.indexToKeyNum(i);
tensorGradient.incrementFeatureByIndex(productFactorValues[i] * multiplier, builderIndex);
}
}
}