/* Copyright (C) 2011 Univ. of Massachusetts Amherst, Computer Science Dept.
This file is part of "MALLET" (MAchine Learning for LanguagE Toolkit).
http://www.cs.umass.edu/~mccallum/mallet
This software is provided under the terms of the Common Public License,
version 1.0, as published by http://www.opensource.org. For further
information, see the file `LICENSE' included with this distribution. */
package cc.mallet.classify.constraints.ge;
import gnu.trove.TDoubleArrayList;
import gnu.trove.TIntArrayList;
import gnu.trove.TIntObjectHashMap;
import java.util.ArrayList;
import java.util.BitSet;
import java.util.HashMap;
import cc.mallet.types.FeatureVector;
import cc.mallet.types.Instance;
import cc.mallet.types.InstanceList;
/**
* Expectation constraint for use with GE.
* Penalizes L_2^2 difference from zero-penalty region [lower,upper].
*
* Multiple constraints are grouped together here
* to make things more efficient.
*
* @author Gregory Druck
*/
public class MaxEntRangeL2FLGEConstraints implements MaxEntGEConstraint {
// maps between input feature indices and constraints
private boolean useValues;
private boolean normalize;
private int numFeatures;
private int numLabels;
protected TIntObjectHashMap<MaxEntL2IndGEConstraint> constraints;
// cache of set of constrained features that fire at last FeatureVector
// provided in preprocess call
protected TIntArrayList indexCache;
protected TDoubleArrayList valueCache;
public MaxEntRangeL2FLGEConstraints(int numFeatures, int numLabels, boolean useValues, boolean normalize) {
this.numFeatures = numFeatures;
this.numLabels = numLabels;
this.useValues = useValues;
this.normalize = normalize;
this.constraints = new TIntObjectHashMap<MaxEntL2IndGEConstraint>();
this.indexCache = new TIntArrayList();
this.valueCache = new TDoubleArrayList();
}
public void addConstraint(int fi, int li, double lower, double upper, double weight) {
if (!constraints.containsKey(fi)) {
constraints.put(fi,new MaxEntL2IndGEConstraint());
}
constraints.get(fi).add(li, lower, upper, weight);
}
public BitSet preProcess(InstanceList data) {
// count
int ii = 0;
int fi;
FeatureVector fv;
BitSet bitSet = new BitSet(data.size());
for (Instance instance : data) {
double weight = data.getInstanceWeight(instance);
fv = (FeatureVector)instance.getData();
for (int loc = 0; loc < fv.numLocations(); loc++) {
fi = fv.indexAtLocation(loc);
if (constraints.containsKey(fi)) {
if (useValues) {
constraints.get(fi).count += weight * fv.valueAtLocation(loc);
}
else {
constraints.get(fi).count += weight;
}
bitSet.set(ii);
}
}
ii++;
// default feature, for label regularization
if (constraints.containsKey(numFeatures)) {
bitSet.set(ii);
constraints.get(numFeatures).count += weight;
}
}
return bitSet;
}
public void preProcess(FeatureVector input) {
indexCache.resetQuick();
if (useValues) valueCache.resetQuick();
int fi;
// cache constrained input features
for (int loc = 0; loc < input.numLocations(); loc++) {
fi = input.indexAtLocation(loc);
if (constraints.containsKey(fi)) {
indexCache.add(fi);
if (useValues) valueCache.add(input.valueAtLocation(loc));
}
}
// default feature, for label regularization
if (constraints.containsKey(numFeatures)) {
indexCache.add(numFeatures);
if (useValues) valueCache.add(1);
}
}
public double getCompositeConstraintFeatureValue(FeatureVector input, int label) {
double value = 0;
for (int i = 0; i < indexCache.size(); i++) {
if (useValues) {
value += constraints.get(indexCache.getQuick(i)).getGradientContribution(label) * valueCache.getQuick(i);
}
else {
value += constraints.get(indexCache.getQuick(i)).getGradientContribution(label);
}
}
return value;
}
public void computeExpectations(FeatureVector input, double[] dist, double weight) {
preProcess(input);
for (int li = 0; li < numLabels; li++) {
double p = weight * dist[li];
for (int i = 0; i < indexCache.size(); i++) {
if (useValues) {
constraints.get(indexCache.getQuick(i)).expectation[li] += p * valueCache.getQuick(i);
}
else {
constraints.get(indexCache.getQuick(i)).expectation[li] += p;
}
}
}
}
public double getValue() {
double value = 0.0;
for (int fi : constraints.keys()) {
MaxEntL2IndGEConstraint constraint = constraints.get(fi);
if ( constraint.count > 0.0) {
// value due to current constraint
for (int labelIndex = 0; labelIndex < numLabels; ++labelIndex) {
value -= constraint.getValue(labelIndex);
}
}
}
assert(!Double.isNaN(value) && !Double.isInfinite(value));
return value;
}
public void zeroExpectations() {
for (int fi : constraints.keys()) {
constraints.get(fi).expectation = new double[constraints.get(fi).getNumConstrainedLabels()];
}
}
protected class MaxEntL2IndGEConstraint {
protected int index;
protected double count;
protected ArrayList<Double> lower;
protected ArrayList<Double> upper;
protected ArrayList<Double> weights;
protected HashMap<Integer,Integer> labelMap;
protected double[] expectation;
public MaxEntL2IndGEConstraint() {
lower = new ArrayList<Double>();
upper = new ArrayList<Double>();
weights = new ArrayList<Double>();
labelMap = new HashMap<Integer,Integer>();
index = 0;
count = 0;
}
public void add(int label, double lower, double upper, double weight) {
this.lower.add(lower);
this.upper.add(upper);
this.weights.add(weight);
labelMap.put(label, index);
index++;
}
public void incrementExpectation(int li, double value) {
if (labelMap.containsKey(li)) {
int i = labelMap.get(li);
expectation[i] += value;
}
}
public double getValue(int li) {
if (labelMap.containsKey(li)) {
int i = labelMap.get(li);
assert(this.count != 0);
double ex;
if (normalize) {
ex = this.expectation[i] / this.count;
}
else {
ex = this.expectation[i];
}
if (ex < lower.get(i)) {
return weights.get(i) * Math.pow(lower.get(i) - ex,2);
}
else if (ex > upper.get(i)) {
return weights.get(i) * Math.pow(upper.get(i) - ex,2);
}
}
return 0;
}
public int getNumConstrainedLabels() {
return index;
}
public double getGradientContribution(int li) {
if (labelMap.containsKey(li)) {
int i = labelMap.get(li);
assert(this.count != 0);
if (normalize) {
double ex = this.expectation[i] / this.count;
if (ex < lower.get(i)) {
return 2 * weights.get(i) * (lower.get(i) / count - expectation[i] / (count * count));
}
else if (ex > upper.get(i)) {
return 2 * weights.get(i) * (upper.get(i) / count - expectation[i] / (count * count));
}
}
else {
double ex = this.expectation[i];
if (ex < lower.get(i)) {
return 2 * weights.get(i) * (lower.get(i) - expectation[i]);
}
else if (ex > upper.get(i)) {
return 2 * weights.get(i) * (upper.get(i) - expectation[i]);
}
}
}
return 0;
}
}
}