Menu
▾
▴
[Jboost-cvs] jboost/src/jboost/booster AbstractBooster.java, 1.7, 1.8 AbstractBoosterTest.java, 1.2, 1.3 AdaBoost.java, 1.5, 1.6 AdaBoostTest.java, 1.1.1.1, 1.2 Bag.java, 1.2, 1.3 BinaryPrediction.java, 1.1.1.1, 1.2 Booster.java, 1.3, 1.4 BoosterTestSuite.java, 1.3, 1.4 BrownBoost.java, 1.10, 1.11 BrownBoostTest.java, 1.3, 1.4 DebugWrap.java, 1.3, 1.4 LogLossBoost.java, 1.2, 1.3 LogLossBoostTest.java, 1.1.1.1, 1.2 MixedBinaryPrediction.java, 1.5, 1.6 MixedBinaryPredictionTest.java, 1.1, 1.2 MulticlassWrapMH.java, 1.5, 1.6 MulticlassWrapOC.java, 1.1.1.1, 1.2 NormalizedPrediction.java, 1.2, 1.3 NotNormalizedPredException.java, 1.1.1.1, 1.2 Prediction.java, 1.2, 1.3 YabaBoost.java, 1.7, 1.8 YabaBoostTest.java, 1.1.1.1, 1.2
Update of /cvsroot/jboost/jboost/src/jboost/booster In directory sc8-pr-cvs17.sourceforge.net:/tmp/cvs-serv8493 Modified Files: AbstractBooster.java AbstractBoosterTest.java AdaBoost.java AdaBoostTest.java Bag.java BinaryPrediction.java Booster.java BoosterTestSuite.java BrownBoost.java BrownBoostTest.java DebugWrap.java LogLossBoost.java LogLossBoostTest.java MixedBinaryPrediction.java MixedBinaryPredictionTest.java MulticlassWrapMH.java MulticlassWrapOC.java NormalizedPrediction.java NotNormalizedPredException.java Prediction.java YabaBoost.java YabaBoostTest.java Log Message: Small comments and sed commands to reformat Index: Bag.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/Bag.java,v retrieving revision 1.2 retrieving revision 1.3 diff -C2 -d -r1.2 -r1.3 *** Bag.java 24 Jan 2008 22:30:16 -0000 1.2 --- Bag.java 25 Mar 2008 01:00:27 -0000 1.3 *************** *** 21,26 **** */ public abstract void reset(); ! ! /** * Get the weights of the bag. */ --- 21,26 ---- */ public abstract void reset(); ! ! /** * Get the weights of the bag. */ *************** *** 28,32 **** return m_w; } ! /** * If the bag has no weight, then it is considered to be weightless. --- 28,32 ---- return m_w; } ! /** * If the bag has no weight, then it is considered to be weightless. *************** *** 106,110 **** * When data splitting is used, this is the same as getLoss() but * computed using only examples in part s of the split. s<0 ! * indicates that the entire dataset should be used. * * The default implementation of this method simply throws an --- 106,110 ---- * When data splitting is used, this is the same as getLoss() but * computed using only examples in part s of the split. s<0 ! * indicates that the entire dataset should be used. * * The default implementation of this method simply throws an *************** *** 114,118 **** throw new NotSupportedException("getLoss",this.getClass().toString()); } ! /* * Computes a non-default loss for this bag. The loss types are --- 114,118 ---- throw new NotSupportedException("getLoss",this.getClass().toString()); } ! /* * Computes a non-default loss for this bag. The loss types are *************** *** 122,126 **** */ /* public abstract double getSpecialLoss(int lossType); */ ! /* * When data splitting is used, this is the same as --- 122,126 ---- */ /* public abstract double getSpecialLoss(int lossType); */ ! /* * When data splitting is used, this is the same as *************** *** 135,139 **** } */ ! /** * Checks if a given loss type is allowed for this bag. --- 135,139 ---- } */ ! /** * Checks if a given loss type is allowed for this bag. Index: MulticlassWrapOC.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/MulticlassWrapOC.java,v retrieving revision 1.1.1.1 retrieving revision 1.2 diff -C2 -d -r1.1.1.1 -r1.2 *** MulticlassWrapOC.java 16 May 2007 04:06:02 -0000 1.1.1.1 --- MulticlassWrapOC.java 25 Mar 2008 01:00:28 -0000 1.2 *************** *** 14,16 **** MulticlassWrapOC(AbstractBooster booster) {} ! } --- 14,16 ---- MulticlassWrapOC(AbstractBooster booster) {} ! } Index: BoosterTestSuite.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/BoosterTestSuite.java,v retrieving revision 1.3 retrieving revision 1.4 diff -C2 -d -r1.3 -r1.4 *** BoosterTestSuite.java 2 Oct 2007 02:32:07 -0000 1.3 --- BoosterTestSuite.java 25 Mar 2008 01:00:27 -0000 1.4 *************** *** 30,33 **** return suite; } ! } --- 30,33 ---- return suite; } ! } Index: AdaBoostTest.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/AdaBoostTest.java,v retrieving revision 1.1.1.1 retrieving revision 1.2 diff -C2 -d -r1.1.1.1 -r1.2 *** AdaBoostTest.java 16 May 2007 04:06:02 -0000 1.1.1.1 --- AdaBoostTest.java 25 Mar 2008 01:00:27 -0000 1.2 *************** *** 74,78 **** //TODO Implement newBag(). } ! public final void testUpdate() { //TODO Implement update(). --- 74,78 ---- //TODO Implement newBag(). } ! public final void testUpdate() { //TODO Implement update(). *************** *** 103,107 **** } } ! for (int i=0; i < ones.length; i++) { ones[i]= indices[COUNT/2 + i]; --- 103,107 ---- } } ! for (int i=0; i < ones.length; i++) { ones[i]= indices[COUNT/2 + i]; *************** *** 116,120 **** } } ! final public void testGetPredictions() { //TODO Implement getPredictions(). --- 116,120 ---- } } ! final public void testGetPredictions() { //TODO Implement getPredictions(). Index: NormalizedPrediction.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/NormalizedPrediction.java,v retrieving revision 1.2 retrieving revision 1.3 diff -C2 -d -r1.2 -r1.3 *** NormalizedPrediction.java 24 Jan 2008 22:30:16 -0000 1.2 --- NormalizedPrediction.java 25 Mar 2008 01:00:28 -0000 1.3 *************** *** 2,6 **** /** ! * A normalized predictor. When a prediction is added, we * ensure that the margin is normalized. This normalization is done * by "mixing" in the new prediction. --- 2,6 ---- /** ! * A normalized predictor. When a prediction is added, we * ensure that the margin is normalized. This normalization is done * by "mixing" in the new prediction. *************** *** 10,14 **** public interface NormalizedPrediction { ! /** * Must be used very carefully. An example: * --- 10,14 ---- public interface NormalizedPrediction { ! /** * Must be used very carefully. An example: * *************** *** 22,28 **** */ public Prediction add(Prediction p); ! /* public static boolean isNormPred(Prediction p); */ ! public double getDeltaT(); } --- 22,28 ---- */ public Prediction add(Prediction p); ! /* public static boolean isNormPred(Prediction p); */ ! public double getDeltaT(); } Index: LogLossBoost.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/LogLossBoost.java,v retrieving revision 1.2 retrieving revision 1.3 diff -C2 -d -r1.2 -r1.3 *** LogLossBoost.java 2 Oct 2007 02:32:07 -0000 1.2 --- LogLossBoost.java 25 Mar 2008 01:00:27 -0000 1.3 *************** *** 6,10 **** * AdaBoost-like implementation for the Log-Loss. The only differnece is weight * update. ! * * @author Yoram Singer * @version $Header: --- 6,10 ---- * AdaBoost-like implementation for the Log-Loss. The only differnece is weight * update. ! * * @author Yoram Singer * @version $Header: *************** *** 17,21 **** /** * Default constructors call AdaBoost constructors ! * Uses a value of 0.0 for the smoothing term */ public LogLossBoost() { --- 17,21 ---- /** * Default constructors call AdaBoost constructors ! * Uses a value of 0.0 for the smoothing term */ public LogLossBoost() { *************** *** 59,63 **** } ! /** * The LogLoss example weight is set to 1/(1 + e^(margin)) --- 59,63 ---- } ! /** * The LogLoss example weight is set to 1/(1 + e^(margin)) *************** *** 67,71 **** } ! /** * Update the examples m_margins and m_weights using the * logistic loss computation --- 67,71 ---- } ! /** * Update the examples m_margins and m_weights using the * logistic loss computation *************** *** 78,82 **** m_oldWeights[i]= m_weights[i]; } ! // update m_weights and m_margins for (int i= 0; i < exampleIndex.length; i++) { --- 78,82 ---- m_oldWeights[i]= m_weights[i]; } ! // update m_weights and m_margins for (int i= 0; i < exampleIndex.length; i++) { Index: BinaryPrediction.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/BinaryPrediction.java,v retrieving revision 1.1.1.1 retrieving revision 1.2 diff -C2 -d -r1.1.1.1 -r1.2 *** BinaryPrediction.java 16 May 2007 04:06:02 -0000 1.1.1.1 --- BinaryPrediction.java 25 Mar 2008 01:00:27 -0000 1.2 *************** *** 13,17 **** public BinaryPrediction(double p) {prediction=p;} public BinaryPrediction() {prediction=0.0;} ! public Object clone(){ Object a = new BinaryPrediction(prediction); --- 13,17 ---- public BinaryPrediction(double p) {prediction=p;} public BinaryPrediction() {prediction=0.0;} ! public Object clone(){ Object a = new BinaryPrediction(prediction); *************** *** 28,37 **** return this; } ! public Prediction add(double w, Prediction p) { prediction += w * ((BinaryPrediction) p).prediction; return this; } ! public double[] getClassScores() { double[] a = {-prediction,prediction}; --- 28,37 ---- return this; } ! public Prediction add(double w, Prediction p) { prediction += w * ((BinaryPrediction) p).prediction; return this; } ! public double[] getClassScores() { double[] a = {-prediction,prediction}; *************** *** 55,61 **** public boolean equals(Prediction other) { BinaryPrediction bp= (BinaryPrediction) other; ! return (prediction == bp.prediction); } ! /** * computes margin as absolute value of prediction. --- 55,61 ---- public boolean equals(Prediction other) { BinaryPrediction bp= (BinaryPrediction) other; ! return (prediction == bp.prediction); } ! /** * computes margin as absolute value of prediction. Index: Booster.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/Booster.java,v retrieving revision 1.3 retrieving revision 1.4 diff -C2 -d -r1.3 -r1.4 *** Booster.java 13 Oct 2007 04:32:28 -0000 1.3 --- Booster.java 25 Mar 2008 01:00:27 -0000 1.4 *************** *** 1,5 **** /* * Created on Jan 12, 2004 ! * */ package jboost.booster; --- 1,5 ---- /* * Created on Jan 12, 2004 ! * */ package jboost.booster; *************** *** 9,13 **** /** ! * This is the definition of the required interface for a booster. * A skeletal implementation is provided by AbstractBooster. * --- 9,13 ---- /** ! * This is the definition of the required interface for a booster. * A skeletal implementation is provided by AbstractBooster. * *************** *** 15,44 **** */ public interface Booster { ! /** * Factory method to create an instance of a Booster * @param configuration ! */ public abstract void init(Configuration config); ! /** * Create and return a new Bag which initially contains the elements in the * list 'defaults'. ! * * @param defaults the initial items to add to the Bag */ public abstract Bag newBag(int[] defaults); ! /** * Create and return a new Bag which is a clone of 'original' ! * * @param original the bag to copy into a new Bag */ public abstract Bag newBag(Bag original); ! /** * Find the best binary split for a sorted list of example indices with given * split points. ! * * @param l an array of example indices, sorted. * @param sp an array with true in position i when a split between positions --- 15,44 ---- */ public interface Booster { ! /** * Factory method to create an instance of a Booster * @param configuration ! */ public abstract void init(Configuration config); ! /** * Create and return a new Bag which initially contains the elements in the * list 'defaults'. ! * * @param defaults the initial items to add to the Bag */ public abstract Bag newBag(int[] defaults); ! /** * Create and return a new Bag which is a clone of 'original' ! * * @param original the bag to copy into a new Bag */ public abstract Bag newBag(Bag original); ! /** * Find the best binary split for a sorted list of example indices with given * split points. ! * * @param l an array of example indices, sorted. * @param sp an array with true in position i when a split between positions *************** *** 50,64 **** */ public abstract int findBestSplit(Bag b0, Bag b1, int[] l, boolean[] sp); ! /** * Compute the loss associated with an array of bags where small loss is * considered "better". The default implementation of this procedure assumes * that the loss is additive across bags. ! * * @param bags array of bags whose losses will be added up and returned * @return loss the sum of the losses for all the bags */ public abstract double getLoss(Bag[] bags); ! /** * Returns a list of margin values for the training data. The actual number --- 50,64 ---- */ public abstract int findBestSplit(Bag b0, Bag b1, int[] l, boolean[] sp); ! /** * Compute the loss associated with an array of bags where small loss is * considered "better". The default implementation of this procedure assumes * that the loss is additive across bags. ! * * @param bags array of bags whose losses will be added up and returned * @return loss the sum of the losses for all the bags */ public abstract double getLoss(Bag[] bags); ! /** * Returns a list of margin values for the training data. The actual number *************** *** 77,86 **** public abstract String getParamString(); ! /** * Returns the current theoretical bound on the training error. */ public abstract double getTheoryBound(); ! /** * Returns the predictions associated with a list of bags representing a --- 77,86 ---- public abstract String getParamString(); ! /** * Returns the current theoretical bound on the training error. */ public abstract double getTheoryBound(); ! /** * Returns the predictions associated with a list of bags representing a *************** *** 94,102 **** */ public abstract Prediction[] getPredictions(Bag[] b, int[][] partition); ! /** * Updates training set data structures to reflect the addition of a new base * classifier. ! * * @param predictions an array of predictions * @param examples an array of arrays of example indexes for which the --- 94,102 ---- */ public abstract Prediction[] getPredictions(Bag[] b, int[][] partition); ! /** * Updates training set data structures to reflect the addition of a new base * classifier. ! * * @param predictions an array of predictions * @param examples an array of arrays of example indexes for which the *************** *** 105,111 **** * predictions</i>. the second index runs over the examples * indices that correspond to that part. ! */ public abstract void update(Prediction[] predictions, int[][] elements); ! /** * Use the margin value to calculate the weight for the example --- 105,111 ---- * predictions</i>. the second index runs over the examples * indices that correspond to that part. ! */ public abstract void update(Prediction[] predictions, int[][] elements); ! /** * Use the margin value to calculate the weight for the example *************** *** 114,128 **** */ public double calculateWeight(double margin); ! /** * Create and return a new Bag which contains no elements */ public abstract Bag newBag(); ! /** * Clear the examples list */ public abstract void clear(); ! /** * Finalize the data structures of the booster. Execute after the constructor --- 114,128 ---- */ public double calculateWeight(double margin); ! /** * Create and return a new Bag which contains no elements */ public abstract Bag newBag(); ! /** * Clear the examples list */ public abstract void clear(); ! /** * Finalize the data structures of the booster. Execute after the constructor *************** *** 130,138 **** */ public abstract void finalizeData(); ! /** * Add an example to location index in the list of examples handled by the * booster. ! * * @param index the position where the example should be added * @param label the label to add --- 130,138 ---- */ public abstract void finalizeData(); ! /** * Add an example to location index in the list of examples handled by the * booster. ! * * @param index the position where the example should be added * @param label the label to add *************** *** 140,144 **** public abstract void addExample(int index, Label label); public abstract void addExample(int index, Label label, double weight); ! } --- 140,144 ---- public abstract void addExample(int index, Label label); public abstract void addExample(int index, Label label, double weight); ! } Index: AdaBoost.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/AdaBoost.java,v retrieving revision 1.5 retrieving revision 1.6 diff -C2 -d -r1.5 -r1.6 *** AdaBoost.java 24 Jan 2008 22:30:16 -0000 1.5 --- AdaBoost.java 25 Mar 2008 01:00:27 -0000 1.6 *************** *** 9,12 **** --- 9,14 ---- import jboost.controller.Configuration; import jboost.examples.Label; + import jboost.booster.BrownBoost; + import jboost.booster.BrownBoost.BrownBag; *************** *** 14,18 **** * The simplest possible implementation of a booster. confidence-rated adaboost * based on equality/inequality of m_labels ! * * @author Yoav Freund * @version $Header: --- 16,20 ---- * The simplest possible implementation of a booster. confidence-rated adaboost * based on equality/inequality of m_labels ! * * @author Yoav Freund * @version $Header: *************** *** 34,37 **** --- 36,46 ---- protected double[] m_potentials; + /** + * The predictions from a hypothesis for an iteration. + * HACK: This should eventually be removed for efficiency concerns. + */ + protected double[] m_hypPredictions; + + /** */ protected int[] m_posExamples; *************** *** 55,59 **** /** if false, then assume all sample weights are 1 */ //protected boolean m_useSampleWeights; ! protected double m_totalWeight; // total weight of all examples protected int m_numExamples= 0; // number of examples in training set --- 64,68 ---- /** if false, then assume all sample weights are 1 */ //protected boolean m_useSampleWeights; ! protected double m_totalWeight; // total weight of all examples protected int m_numExamples= 0; // number of examples in training set *************** *** 67,72 **** ! /** ! * default constructor * */ --- 76,81 ---- ! /** ! * default constructor * */ *************** *** 77,81 **** /** * Constructor which takes a smoothing factor ! * * @param smooth "smoothing" factor */ --- 86,90 ---- /** * Constructor which takes a smoothing factor ! * * @param smooth "smoothing" factor */ *************** *** 94,100 **** m_smooth= config.getDouble(PREFIX + "smooth", 0.5); } ! /** ! * Add an example to the data set of this booster * @param index * @param label --- 103,109 ---- m_smooth= config.getDouble(PREFIX + "smooth", 0.5); } ! /** ! * Add an example to the data set of this booster * @param index * @param label *************** *** 114,118 **** } } else { ! failed= "Adaboost.addExample expected a label which is either 0 or 1. It received " + l; } --- 123,127 ---- } } else { ! failed= "Adaboost.addExample expected a label which is either 0 or 1. It received " + l; } *************** *** 122,129 **** } } ! /** * Add an example to the dataset ! * Default the weight for this example to 1 * If this method is used, then this booster will assume * that all the sample weights are 1 --- 131,138 ---- } } ! /** * Add an example to the dataset ! * Default the weight for this example to 1 * If this method is used, then this booster will assume * that all the sample weights are 1 *************** *** 167,179 **** m_weights[index]= m_oldWeights[index]= defaultWeight; m_labels[index]= a.getLabel(); ! if (a.getLabel()==POSITIVE_LABEL) ! m_posExamples[m_posIndex++] = index; else if (a.getLabel()==NEGATIVE_LABEL) ! m_negExamples[m_negIndex++] = index; else { System.err.println("Label of example is unknown to adaboost"); System.exit(2); } ! m_sampleWeights[index]= a.getWeight(); } --- 176,188 ---- m_weights[index]= m_oldWeights[index]= defaultWeight; m_labels[index]= a.getLabel(); ! if (a.getLabel()==POSITIVE_LABEL) ! m_posExamples[m_posIndex++] = index; else if (a.getLabel()==NEGATIVE_LABEL) ! m_negExamples[m_negIndex++] = index; else { System.err.println("Label of example is unknown to adaboost"); System.exit(2); } ! m_sampleWeights[index]= a.getWeight(); } *************** *** 181,185 **** m_tmpList.clear(); // free the memory } ! public void finalizeData() { finalizeData(1.0); --- 190,194 ---- m_tmpList.clear(); // free the memory } ! public void finalizeData() { finalizeData(1.0); *************** *** 213,217 **** return r; } ! /** * --- 222,226 ---- return r; } ! /** * *************** *** 230,234 **** return m_numExamples; } ! /** * --- 239,243 ---- return m_numExamples; } ! /** * *************** *** 237,242 **** return m_totalWeight; } ! ! /** * Returns a string with all the weights, margins, etc --- 246,251 ---- return m_totalWeight; } ! ! /** * Returns a string with all the weights, margins, etc *************** *** 246,251 **** ret.append(getParamString()); for (int i=0; i<m_margins.length; i++){ ! ret.append(String.format("[%d];[%.4f];[%.4f];[%.4f];\n", ! m_labels[i], m_margins[i], m_weights[i], m_potentials[i])); } --- 255,260 ---- ret.append(getParamString()); for (int i=0; i<m_margins.length; i++){ ! ret.append(String.format("[%d];[%.4f];[%.4f];[%.4f];\n", ! m_labels[i], m_margins[i], m_weights[i], m_potentials[i])); } *************** *** 318,321 **** --- 327,426 ---- + + /** + * @param z - any double + * @return If z is negative return -1 else return 1 + */ + public static double sign(double z){ + if( Double.compare(z, 0.0) == 0 ){ + return 1.0; + } else if( Double.compare(z, -0.0) == 0 ){ + return -1.0; + } + + if(z > 0){ + return 1.0; + } else { + return -1.0; + } + } + + /** + * Get the "step" of the hypothesis on an example. The step is + * defined as $y_j * h_i(x_j)$ for example j and hypothesis i. + * @param simple_label - The label of the example as given by m_labels + * @param hyp_pred - The hypothesized value of the example + * @return +1 if label matches hyp, -1 if label doesn't match hyp, 0 if no hyp + */ + public double getStep(short simple_label, double hyp_pred) { + double step = getLabel(simple_label)*hyp_pred; + double EPS = 0.000001; + if(Math.abs(step) < EPS) return 0.0; + return sign(step); + } + + public double getLabel(short simple_label) { + return sign(-simple_label+0.5); + } + + protected double getHypErr(Bag[] bags, int[][] exampleIndex) { + double hyp_err = 0.0; + double gamma = 0.0; + double total_weight = 0.0; + + // Keep track of which hypotheses had hypotheses associated with them. + boolean[] examplesWithHyp = new boolean[m_margins.length]; + m_hypPredictions = new double[m_margins.length]; + for (int i=0; i < exampleIndex.length; i++){ + int[] index = exampleIndex[i]; + BinaryBag b = (BinaryBag) bags[i]; + for (int j=0; j < index.length; j++){ + int example = index[j]; + if (this instanceof BrownBoost) + m_hypPredictions[example] = ((BrownBag)b).calcPrediction(1.0,1.0).getClassScores()[0]; + else + m_hypPredictions[example] = b.calcPrediction().getClassScores()[0]; + } + } + + + int numExamplesWithHyps = 0; + double weight; + // Get all examples that have a hypothesis associated with them + for (int i= 0; i < exampleIndex.length; i++) { + int[] indexes= exampleIndex[i]; + for (int j= 0; j < indexes.length; j++) { + int example = indexes[j]; + examplesWithHyp[example] = true; + numExamplesWithHyps += 1; + + double step = getStep(m_labels[example], m_hypPredictions[example]); + gamma += m_weights[example]*step; + if (step < 0) // We got it wrong + hyp_err += 1; + } + } + + // Get all examples that have no hypothesis associated with them. + for (int i=0; i < m_margins.length; i++) { + total_weight += m_weights[i]; + if(!examplesWithHyp[i]){ + m_hypPredictions[i] = 0; + //System.out.println("m_hypPredictions[" + i + "," + example + "]: " + 0 + " (No hyp for example " + example + ")"); + } + } + + hyp_err /= numExamplesWithHyps; + gamma /= (double)total_weight; + + if (numExamplesWithHyps > 0) { + System.out.println("Num Examples with predictions: " + numExamplesWithHyps + "/" + m_margins.length); + System.out.println("Gamma: " + gamma); + } + return gamma; + } + + + /** * Returns the predictions associated with a list of bags representing a *************** *** 337,345 **** } }*/ return getPredictions(bags); } ! ! /** * AdaBoost uses e^(-margin) as the weight calculation --- 442,451 ---- } }*/ + double gamma = getHypErr(bags, exampleIndex); return getPredictions(bags); } ! ! /** * AdaBoost uses e^(-margin) as the weight calculation *************** *** 348,358 **** return Math.exp(-1 * margin); } ! /** ! * Update the examples m_margins and m_weights using the * exponential update * @param predictions values for examples * @param exampleIndex the list of examples to update ! */ public void update(Prediction[] predictions, int[][] exampleIndex) { // save old m_weights --- 454,464 ---- return Math.exp(-1 * margin); } ! /** ! * Update the examples m_margins and m_weights using the * exponential update * @param predictions values for examples * @param exampleIndex the list of examples to update ! */ public void update(Prediction[] predictions, int[][] exampleIndex) { // save old m_weights *************** *** 375,380 **** } ! /** ! * Defines the state of an example * Inner class used to store a list of Examples * The list is converted into the internal data structures for the --- 481,486 ---- } ! /** ! * Defines the state of an example * Inner class used to store a list of Examples * The list is converted into the internal data structures for the *************** *** 385,389 **** short m_label; double m_weight; ! /** * Ctor for a TmpData object --- 491,495 ---- short m_label; double m_weight; ! /** * Ctor for a TmpData object *************** *** 397,401 **** m_weight= weight; } ! /** * Get the index for this example --- 503,507 ---- m_weight= weight; } ! /** * Get the index for this example *************** *** 405,409 **** return m_index; } ! /** * Get the label for this example --- 511,515 ---- return m_index; } ! /** * Get the label for this example *************** *** 413,417 **** return m_label; } ! /** * Get the weigh for this example --- 519,523 ---- return m_label; } ! /** * Get the weigh for this example *************** *** 423,427 **** } ! /** * This is the definition of a bag for AdaBoost. The two m_labels are --- 529,533 ---- } ! /** * This is the definition of a bag for AdaBoost. The two m_labels are *************** *** 482,487 **** * Update the weights in this bag using the weights from the booster * The example index is used to find the label and weight for this example ! * ! * @param index the example that is being added to this bag. The index refers to the booster's * internal data structures */ --- 588,593 ---- * Update the weights in this bag using the weights from the booster * The example index is used to find the label and weight for this example ! * ! * @param index the example that is being added to this bag. The index refers to the booster's * internal data structures */ *************** *** 553,561 **** double smoothFactor= m_epsilon * m_totalWeight; double EPS = 1e-50; ! if (Double.isNaN(smoothFactor) || (Math.abs(m_totalWeight)<EPS) || (Math.abs(smoothFactor)<EPS) || Double.isNaN(m_totalWeight)) { return new BinaryPrediction(0.0); } ! BinaryPrediction p = new BinaryPrediction( m_w[1] == m_w[0] --- 659,667 ---- double smoothFactor= m_epsilon * m_totalWeight; double EPS = 1e-50; ! if (Double.isNaN(smoothFactor) || (Math.abs(m_totalWeight)<EPS) || (Math.abs(smoothFactor)<EPS) || Double.isNaN(m_totalWeight)) { return new BinaryPrediction(0.0); } ! BinaryPrediction p = new BinaryPrediction( m_w[1] == m_w[0] *************** *** 568,572 **** /** * Compare a bag to this bag and output true if they are equal ! * * @param other * bag to compare to this bag --- 674,678 ---- /** * Compare a bag to this bag and output true if they are equal ! * * @param other * bag to compare to this bag Index: NotNormalizedPredException.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/NotNormalizedPredException.java,v retrieving revision 1.1.1.1 retrieving revision 1.2 diff -C2 -d -r1.1.1.1 -r1.2 *** NotNormalizedPredException.java 16 May 2007 04:06:02 -0000 1.1.1.1 --- NotNormalizedPredException.java 25 Mar 2008 01:00:28 -0000 1.2 *************** *** 1,11 **** package jboost.booster; ! /** * This exception gets thrown if a prediction that is supposed to be ! * normalized, becomes unnormalized. * @author Aaron Arvey */ public class NotNormalizedPredException extends RuntimeException { ! NotNormalizedPredException (String m) { message = m; } --- 1,11 ---- package jboost.booster; ! /** * This exception gets thrown if a prediction that is supposed to be ! * normalized, becomes unnormalized. * @author Aaron Arvey */ public class NotNormalizedPredException extends RuntimeException { ! NotNormalizedPredException (String m) { message = m; } Index: BrownBoostTest.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/BrownBoostTest.java,v retrieving revision 1.3 retrieving revision 1.4 diff -C2 -d -r1.3 -r1.4 *** BrownBoostTest.java 10 Jul 2007 01:32:52 -0000 1.3 --- BrownBoostTest.java 25 Mar 2008 01:00:27 -0000 1.4 *************** *** 76,83 **** //TODO Implement newBag(). } ! ! /** ! * This test indicates whether or not the update algorithm in * BrownBoost is producing the correct weights. This is the core * of the BrownBoost algorithm and is thus the most important to test. --- 76,83 ---- //TODO Implement newBag(). } ! ! /** ! * This test indicates whether or not the update algorithm in * BrownBoost is producing the correct weights. This is the core * of the BrownBoost algorithm and is thus the most important to test. *************** *** 123,127 **** } } ! for (int i=0; i < ones.length; i++) { ones[i]= indices[COUNT/2 + i]; --- 123,127 ---- } } ! for (int i=0; i < ones.length; i++) { ones[i]= indices[COUNT/2 + i]; *************** *** 146,150 **** */ } ! final public void testGetPredictions() { //TODO Implement getPredictions(). --- 146,150 ---- */ } ! final public void testGetPredictions() { //TODO Implement getPredictions(). Index: LogLossBoostTest.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/LogLossBoostTest.java,v retrieving revision 1.1.1.1 retrieving revision 1.2 diff -C2 -d -r1.1.1.1 -r1.2 *** LogLossBoostTest.java 16 May 2007 04:06:02 -0000 1.1.1.1 --- LogLossBoostTest.java 25 Mar 2008 01:00:27 -0000 1.2 *************** *** 10,16 **** */ public class LogLossBoostTest extends AbstractBoosterTest { ! LogLossBoost m_logLoss; ! /** * @param arg0 --- 10,16 ---- */ public class LogLossBoostTest extends AbstractBoosterTest { ! LogLossBoost m_logLoss; ! /** * @param arg0 *************** *** 37,42 **** m_logLoss= new LogLossBoost(); } ! ! public final void testFinalizeData() { //TODO Implement finalizeData(). --- 37,42 ---- m_logLoss= new LogLossBoost(); } ! ! public final void testFinalizeData() { //TODO Implement finalizeData(). *************** *** 71,75 **** } } ! for (int i=0; i < ones.length; i++) { ones[i]= indices[COUNT/2 + i]; --- 71,75 ---- } } ! for (int i=0; i < ones.length; i++) { ones[i]= indices[COUNT/2 + i]; Index: MixedBinaryPrediction.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/MixedBinaryPrediction.java,v retrieving revision 1.5 retrieving revision 1.6 diff -C2 -d -r1.5 -r1.6 *** MixedBinaryPrediction.java 24 Jan 2008 22:30:16 -0000 1.5 --- MixedBinaryPrediction.java 25 Mar 2008 01:00:27 -0000 1.6 *************** *** 16,20 **** public MixedBinaryPrediction() {super();} ! public MixedBinaryPrediction(double p) throws NotNormalizedPredException { (new Exception("asdf")).printStackTrace(); --- 16,20 ---- public MixedBinaryPrediction() {super();} ! public MixedBinaryPrediction(double p) throws NotNormalizedPredException { (new Exception("asdf")).printStackTrace(); *************** *** 28,32 **** } ! public MixedBinaryPrediction(double p, double dt, double nc) throws NotNormalizedPredException { super(); --- 28,32 ---- } ! public MixedBinaryPrediction(double p, double dt, double nc) throws NotNormalizedPredException { super(); *************** *** 35,41 **** normalizingConstant = nc; } - ! public MixedBinaryPrediction(double p, double dt) throws NotNormalizedPredException { super(); --- 35,41 ---- normalizingConstant = nc; } ! ! public MixedBinaryPrediction(double p, double dt) throws NotNormalizedPredException { super(); *************** *** 44,48 **** normalizingConstant = 1; } ! --- 44,48 ---- normalizingConstant = 1; } ! *************** *** 52,56 **** } ! /** * Be very careful with how this is used. See NormalizedPrediction * for details. --- 52,56 ---- } ! /** * Be very careful with how this is used. See NormalizedPrediction * for details. *************** *** 94,98 **** return normalizingConstant; } ! public Prediction add(double w, Prediction p) { ((MixedBinaryPrediction) p).scale(w); --- 94,98 ---- return normalizingConstant; } ! public Prediction add(double w, Prediction p) { ((MixedBinaryPrediction) p).scale(w); *************** *** 100,109 **** return this; } ! public boolean equals(Prediction other) { MixedBinaryPrediction bp= (MixedBinaryPrediction) other; ! return (prediction == bp.prediction && deltaT == bp.deltaT && normalizingConstant == bp.normalizingConstant); } ! public String toString() { return "MixedBinaryPrediction. p(1)= " + prediction; --- 100,109 ---- return this; } ! public boolean equals(Prediction other) { MixedBinaryPrediction bp= (MixedBinaryPrediction) other; ! return (prediction == bp.prediction && deltaT == bp.deltaT && normalizingConstant == bp.normalizingConstant); } ! public String toString() { return "MixedBinaryPrediction. p(1)= " + prediction; Index: MulticlassWrapMH.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/MulticlassWrapMH.java,v retrieving revision 1.5 retrieving revision 1.6 diff -C2 -d -r1.5 -r1.6 *** MulticlassWrapMH.java 13 Oct 2007 04:32:28 -0000 1.5 --- MulticlassWrapMH.java 25 Mar 2008 01:00:28 -0000 1.6 *************** *** 11,15 **** * no knowledge of the surrounding wrapper. * ! * The multiclass problem is reduce to a binary problem using a * "one-against-all" reduction. This reduction maps a multiclass * example to k (the # of classes) binary examples. Namely, example --- 11,15 ---- * no knowledge of the surrounding wrapper. * ! * The multiclass problem is reduce to a binary problem using a * "one-against-all" reduction. This reduction maps a multiclass * example to k (the # of classes) binary examples. Namely, example *************** *** 25,36 **** /** The underlying m_booster */ ! private AbstractBooster m_booster; /** The number of labels */ ! private int m_numLabels; ! /** The number of labels */ ! private boolean m_isMultiLabel; ! /** * constructor --- 25,36 ---- /** The underlying m_booster */ ! private AbstractBooster m_booster; /** The number of labels */ ! private int m_numLabels; ! /** The number of labels */ ! private boolean m_isMultiLabel; ! /** * constructor *************** *** 43,47 **** m_isMultiLabel = isMultiLabel; } ! public String toString() { if (m_booster==null) { --- 43,47 ---- m_isMultiLabel = isMultiLabel; } ! public String toString() { if (m_booster==null) { *************** *** 52,64 **** System.err.println(msg); } ! return ("MulticlassWrapMH. # of classes = " + m_numLabels + ".\nUnderlying m_booster:\n" + m_booster); } ! public void addExample(int index, Label l) { addExample(index, l, 1.0); } ! /** * @see jboost.booster.Booster#addExample(int, jboost.examples.Label, double) --- 52,64 ---- System.err.println(msg); } ! return ("MulticlassWrapMH. # of classes = " + m_numLabels + ".\nUnderlying m_booster:\n" + m_booster); } ! public void addExample(int index, Label l) { addExample(index, l, 1.0); } ! /** * @see jboost.booster.Booster#addExample(int, jboost.examples.Label, double) *************** *** 80,88 **** m_booster.finalizeData(); } ! public void clear() { m_booster.clear(); } ! public Bag newBag() { return new MultiBag(); --- 80,88 ---- m_booster.finalizeData(); } ! public void clear() { m_booster.clear(); } ! public Bag newBag() { return new MultiBag(); *************** *** 98,102 **** double[][] weights = m_booster.getWeights(); for (int i= 0; i < numExamples; i++) ! for (int j=0; j < m_numLabels; j++) r[i][j]= weights[i*m_numLabels+j][0]; return r; --- 98,102 ---- double[][] weights = m_booster.getWeights(); for (int i= 0; i < numExamples; i++) ! for (int j=0; j < m_numLabels; j++) r[i][j]= weights[i*m_numLabels+j][0]; return r; *************** *** 126,130 **** ! /** * According to Rob: "This method should never be called. We need --- 126,130 ---- ! /** * According to Rob: "This method should never be called. We need *************** *** 142,147 **** return m_booster.calculateWeight(margin); } ! ! public void update(Prediction[] preds, int[][] index) { int num_preds = preds.length; --- 142,147 ---- return m_booster.calculateWeight(margin); } ! ! public void update(Prediction[] preds, int[][] index) { int num_preds = preds.length; *************** *** 149,153 **** int[][] uindex = new int[index.length * m_numLabels][]; int i, j, t, k; ! // create array of predictions to pass to underlying booster --- 149,153 ---- int[][] uindex = new int[index.length * m_numLabels][]; int i, j, t, k; ! // create array of predictions to pass to underlying booster *************** *** 159,163 **** } } ! // create array of indices to pass to underlying booster t = 0; --- 159,163 ---- } } ! // create array of indices to pass to underlying booster t = 0; *************** *** 172,176 **** m_booster.update(upreds, uindex); } ! /** * computes theoretical bound as (m_numLabels/2) * theoretical --- 172,176 ---- m_booster.update(upreds, uindex); } ! /** * computes theoretical bound as (m_numLabels/2) * theoretical *************** *** 186,190 **** return m_booster.getTotalWeight(); } ! /** * returns the margin values of all of the "examples" used to --- 186,190 ---- return m_booster.getTotalWeight(); } ! /** * returns the margin values of all of the "examples" used to *************** *** 194,198 **** return m_booster.getMargins(); } ! /** * Returns the predictions associated with a list of bags representing a --- 194,198 ---- return m_booster.getMargins(); } ! /** * Returns the predictions associated with a list of bags representing a *************** *** 201,220 **** public Prediction[] getPredictions(Bag[] b, int[][] exampleIndex) { Bag[] ubags = new Bag[b.length * m_numLabels]; ! for (int i = 0; i < b.length; i++) for (int j = 0; j < m_numLabels; j++) ubags[i * m_numLabels + j] = ((MultiBag) b[i]).bags[j]; ! Prediction[] upreds = m_booster.getPredictions(ubags, exampleIndex); Prediction[] preds = new Prediction[b.length]; ! for (int i = 0; i < b.length; i++) { preds[i] = new MultiPrediction(); for (int j = 0; j < m_numLabels; j++) ! ((MultiPrediction) preds[i]).preds[j] = upreds[i * m_numLabels + j]; } return preds; ! } public Prediction[] getPredictions(Bag[] b) { --- 201,220 ---- public Prediction[] getPredictions(Bag[] b, int[][] exampleIndex) { Bag[] ubags = new Bag[b.length * m_numLabels]; ! for (int i = 0; i < b.length; i++) for (int j = 0; j < m_numLabels; j++) ubags[i * m_numLabels + j] = ((MultiBag) b[i]).bags[j]; ! Prediction[] upreds = m_booster.getPredictions(ubags, exampleIndex); Prediction[] preds = new Prediction[b.length]; ! for (int i = 0; i < b.length; i++) { preds[i] = new MultiPrediction(); for (int j = 0; j < m_numLabels; j++) ! ((MultiPrediction) preds[i]).preds[j] = upreds[i * m_numLabels + j]; } return preds; ! } public Prediction[] getPredictions(Bag[] b) { *************** *** 223,238 **** return nothing; } ! ! public double getLoss(Bag[] b) { Bag[] ubags = new Bag[b.length * m_numLabels]; ! for (int i = 0; i < b.length; i++) for (int j = 0; j < m_numLabels; j++) ubags[i * m_numLabels + j] = ((MultiBag) b[i]).bags[j]; ! return m_booster.getLoss(ubags); } ! /** * This is the bag class associated with this booster. Each bag --- 223,238 ---- return nothing; } ! ! public double getLoss(Bag[] b) { Bag[] ubags = new Bag[b.length * m_numLabels]; ! for (int i = 0; i < b.length; i++) for (int j = 0; j < m_numLabels; j++) ubags[i * m_numLabels + j] = ((MultiBag) b[i]).bags[j]; ! return m_booster.getLoss(ubags); } ! /** * This is the bag class associated with this booster. Each bag *************** *** 241,247 **** */ class MultiBag extends Bag { ! private Bag[] bags; // underlying bags ! private MultiBag() { bags = new Bag[m_numLabels]; --- 241,247 ---- */ class MultiBag extends Bag { ! private Bag[] bags; // underlying bags ! private MultiBag() { bags = new Bag[m_numLabels]; *************** *** 249,253 **** bags[j] = m_booster.newBag(); } ! public String toString() { String s = "MultiBag.\n"; --- 249,253 ---- bags[j] = m_booster.newBag(); } ! public String toString() { String s = "MultiBag.\n"; *************** *** 256,265 **** return s; } ! public void reset() { for (int j = 0; j < m_numLabels; j++) bags[j].reset(); } ! public boolean isWeightless() { for (int j = 0; j < m_numLabels; j++) --- 256,265 ---- return s; } ! public void reset() { for (int j = 0; j < m_numLabels; j++) bags[j].reset(); } ! public boolean isWeightless() { for (int j = 0; j < m_numLabels; j++) *************** *** 275,279 **** bags[j].addExample(s + j); } ! public void subtractExample(int index) { int s = index * m_numLabels; --- 275,279 ---- bags[j].addExample(s + j); } ! public void subtractExample(int index) { int s = index * m_numLabels; *************** *** 281,285 **** bags[j].subtractExample(s + j); } ! public void addExampleList(int[] l) { int i; --- 281,285 ---- bags[j].subtractExample(s + j); } ! public void addExampleList(int[] l) { int i; *************** *** 294,298 **** } } ! public void subtractExampleList(int[] l) { int i; --- 294,298 ---- } } ! public void subtractExampleList(int[] l) { int i; *************** *** 307,311 **** } } ! public void addBag(Bag b) { MultiBag other = (MultiBag) b; --- 307,311 ---- } } ! public void addBag(Bag b) { MultiBag other = (MultiBag) b; *************** *** 313,317 **** bags[j].addBag(other.bags[j]); } ! public void subtractBag(Bag b) { MultiBag other = (MultiBag) b; --- 313,317 ---- bags[j].addBag(other.bags[j]); } ! public void subtractBag(Bag b) { MultiBag other = (MultiBag) b; *************** *** 319,323 **** bags[j].subtractBag(other.bags[j]); } ! public void copyBag(Bag b) { MultiBag other = (MultiBag) b; --- 319,323 ---- bags[j].subtractBag(other.bags[j]); } ! public void copyBag(Bag b) { MultiBag other = (MultiBag) b; *************** *** 325,329 **** bags[j].copyBag(other.bags[j]); } ! public void refresh(int index) { int s = index * m_numLabels; --- 325,329 ---- bags[j].copyBag(other.bags[j]); } ! public void refresh(int index) { int s = index * m_numLabels; *************** *** 331,335 **** bags[j].refresh(s + j); } ! public void refreshList(int[] l) { int i; --- 331,335 ---- bags[j].refresh(s + j); } ! public void refreshList(int[] l) { int i; *************** *** 344,348 **** } } ! /** * Computes the loss for this bag. This loss is only meaningful for --- 344,348 ---- } } ! /** * Computes the loss for this bag. This loss is only meaningful for *************** *** 355,359 **** return loss; } ! public double getLoss(int s) throws jboost.NotSupportedException { --- 355,359 ---- return loss; } ! public double getLoss(int s) throws jboost.NotSupportedException { *************** *** 364,368 **** } } ! --- 364,368 ---- } } ! *************** *** 370,374 **** * This is the prediction class associated with this booster. * Each prediction is composed of an array of predictions from the ! * underlying booster, one for each class. */ class MultiPrediction extends Prediction { --- 370,374 ---- * This is the prediction class associated with this booster. * Each prediction is composed of an array of predictions from the ! * underlying booster, one for each class. */ class MultiPrediction extends Prediction { *************** *** 377,381 **** */ private Prediction[] preds; ! /** * Constructor. --- 377,381 ---- */ private Prediction[] preds; ! /** * Constructor. *************** *** 384,391 **** preds = new Prediction[m_numLabels]; } ! public Object clone() { MultiPrediction newpred = new MultiPrediction(); ! for (int j = 0; j < m_numLabels; j++) { newpred.preds[j] = (Prediction) preds[j].clone(); --- 384,391 ---- preds = new Prediction[m_numLabels]; } ! public Object clone() { MultiPrediction newpred = new MultiPrediction(); ! for (int j = 0; j < m_numLabels; j++) { newpred.preds[j] = (Prediction) preds[j].clone(); *************** *** 393,397 **** return newpred; } ! public Prediction add(Prediction p) { for (int j = 0; j < m_numLabels; j++) { --- 393,397 ---- return newpred; } ! public Prediction add(Prediction p) { for (int j = 0; j < m_numLabels; j++) { *************** *** 400,404 **** return this; } ! public Prediction scale(double w) { for (int j = 0; j < m_numLabels; j++) --- 400,404 ---- return this; } ! public Prediction scale(double w) { for (int j = 0; j < m_numLabels; j++) *************** *** 406,410 **** return this; } ! public Prediction add(double w, Prediction p) { for (int j = 0; j < m_numLabels; j++) { --- 406,410 ---- return this; } ! public Prediction add(double w, Prediction p) { for (int j = 0; j < m_numLabels; j++) { *************** *** 413,417 **** return this; } ! public double[] getMarginsSingleLabel(Label l) { int maxClass = -1; --- 413,417 ---- return this; } ! public double[] getMarginsSingleLabel(Label l) { int maxClass = -1; *************** *** 433,437 **** return ret; } ! public double[] getMarginsMultiLabel(Label l) { double[] ret = new double[m_numLabels]; --- 433,437 ---- return ret; } ! public double[] getMarginsMultiLabel(Label l) { double[] ret = new double[m_numLabels]; *************** *** 444,457 **** public double[] getMargins(Label l) { ! if (m_isMultiLabel) return getMarginsMultiLabel(l); return getMarginsSingleLabel(l); } ! public double[] getClassScores() { double[] scores = new double[m_numLabels]; double[] uscore; ! for (int j = 0; j < m_numLabels; j++) { uscore = preds[j].getClassScores(); --- 444,457 ---- public double[] getMargins(Label l) { ! if (m_isMultiLabel) return getMarginsMultiLabel(l); return getMarginsSingleLabel(l); } ! public double[] getClassScores() { double[] scores = new double[m_numLabels]; double[] uscore; ! for (int j = 0; j < m_numLabels; j++) { uscore = preds[j].getClassScores(); *************** *** 460,464 **** return scores; } ! /** * Check to see if this MultiPrediction is the same as the other --- 460,464 ---- return scores; } ! /** * Check to see if this MultiPrediction is the same as the other *************** *** 476,480 **** return retval; } ! public String toString() { String s = "MultiPrediction.\n"; --- 476,480 ---- return retval; } ! public String toString() { String s = "MultiPrediction.\n"; *************** *** 483,487 **** return s; } ! public String shortText() { String s = "[,"+preds[0]; --- 483,487 ---- return s; } ! public String shortText() { String s = "[,"+preds[0]; *************** *** 490,504 **** return s+"]"; } ! public String cPreamble() { String code = ""; ! code += "typedef double Prediction_t[" + m_numLabels + "];\n"; ! code += "#define reset_pred() { \\\n"; for (int i = 0; i < m_numLabels; i++) code += " p["+i+"] = 0.0; \\\n"; code += " }\n"; ! code += "#define add_pred("; for (int i = 0; i < m_numLabels; i++) --- 490,504 ---- return s+"]"; } ! public String cPreamble() { String code = ""; ! code += "typedef double Prediction_t[" + m_numLabels + "];\n"; ! code += "#define reset_pred() { \\\n"; for (int i = 0; i < m_numLabels; i++) code += " p["+i+"] = 0.0; \\\n"; code += " }\n"; ! code += "#define add_pred("; for (int i = 0; i < m_numLabels; i++) *************** *** 508,512 **** code += " p["+i+"] += X"+i+"; \\\n"; code += " }\n"; ! code += "#define finalize_pred() \\\n"; code += " (r ? ( \\\n"; --- 508,512 ---- code += " p["+i+"] += X"+i+"; \\\n"; code += " }\n"; ! code += "#define finalize_pred() \\\n"; code += " (r ? ( \\\n"; *************** *** 514,524 **** code += " r["+i+"] = p["+i+"], \\\n"; code += " p[0]) : p[0])\n"; ! return code; } ! public String javaPreamble() { String code = ""; ! code += "" + " static private double[] p = new double[" + m_numLabels + "];\n" --- 514,524 ---- code += " r["+i+"] = p["+i+"], \\\n"; code += " p[0]) : p[0])\n"; ! return code; } ! public String javaPreamble() { String code = ""; ! code += "" + " static private double[] p = new double[" + m_numLabels + "];\n" *************** *** 539,548 **** return code; } ! public double[] toCodeArray() { return getClassScores(); } } ! /** a main for testing */ public static void main(String[] argv) { --- 539,548 ---- return code; } ! public double[] toCodeArray() { return getClassScores(); } } ! /** a main for testing */ public static void main(String[] argv) { *************** *** 550,561 **** AbstractBooster ada = new DebugWrap(new MulticlassWrapMH(new DebugWrap(new AdaBoost(0.0)), 2, true)); ! for(int i=0; i< 10; i++) { ada.addExample(i,new Label(i % 2)); } ! ada.finalizeData(); if(Monitor.logLevel>3) Monitor.log(ada); ! int[] elements = {1,2,6,3,4}; String s="\n Generating a bag with elements:\n" +elements[0]; --- 550,561 ---- AbstractBooster ada = new DebugWrap(new MulticlassWrapMH(new DebugWrap(new AdaBoost(0.0)), 2, true)); ! for(int i=0; i< 10; i++) { ada.addExample(i,new Label(i % 2)); } ! ada.finalizeData(); if(Monitor.logLevel>3) Monitor.log(ada); ! int[] elements = {1,2,6,3,4}; String s="\n Generating a bag with elements:\n" +elements[0]; *************** *** 564,573 **** Bag bag = ada.newBag(elements); if(Monitor.logLevel>3) Monitor.log(bag); ! Prediction[] p; p = ada.getPredictions(new Bag[] {bag}); // calc optimal prediction for this bag ! if(Monitor.logLevel>3) Monitor.log("best prediction for this bag is " + p[0].getClassScores()[1]); ! int[][] exampleList = new int[1][]; exampleList[0] = elements; --- 564,573 ---- Bag bag = ada.newBag(elements); if(Monitor.logLevel>3) Monitor.log(bag); ! Prediction[] p; p = ada.getPredictions(new Bag[] {bag}); // calc optimal prediction for this bag ! if(Monitor.logLevel>3) Monitor.log("best prediction for this bag is " + p[0].getClassScores()[1]); ! int[][] exampleList = new int[1][]; exampleList[0] = elements; *************** *** 579,583 **** if(Monitor.logLevel>3) Monitor.log("now best prediction for this bag is "+ (ada.getPredictions(new Bag[] {bag})[0]).getClassScores()[1]); ! } catch(Exception e) { --- 579,583 ---- if(Monitor.logLevel>3) Monitor.log("now best prediction for this bag is "+ (ada.getPredictions(new Bag[] {bag})[0]).getClassScores()[1]); ! } catch(Exception e) { *************** *** 585,592 **** e.printStackTrace(); } ! ! } ! /* (non-Javadoc) * @see jboost.booster.Booster#init(jboost.controller.Configuration) --- 585,592 ---- e.printStackTrace(); } ! ! } ! /* (non-Javadoc) * @see jboost.booster.Booster#init(jboost.controller.Configuration) *************** *** 594,598 **** public void init(Configuration config) { // TODO Auto-generated method stub ! } } --- 594,598 ---- public void init(Configuration config) { // TODO Auto-generated method stub ! } } Index: MixedBinaryPredictionTest.java =================================================================== RCS file: /cvsroot/jboost/jboost/src/jboost/booster/MixedBinaryPredictionTest.java,v retrieving revision 1.1 retrieving revision 1.2 diff -C2 -d -r1.1 -r1.2 *** MixedBinaryPredictionTest.java 2 Oct 2007 02:32:48 -0000 1.1 --- MixedBinaryPredictionTest.java 25 Mar 2008 01:00:27 -0000 1.2 *************** *** 15,19 **** /** ! * Constructor for * @param arg0 */ --- 15,19 ---- /** ! * Constructor for * @param arg0 */ *************** *** 22,26 **** System.out.println("Making MixedPred class"); } ! final public void testAdd() { System.out.println("test add"); --- 22,26 ---- System.out.println("Making MixedPred class"); } ! final public void testAdd() { System.out.println("test add"); *************** *** 29,33 **** MixedBinaryPrediction np = new MixedBinaryPrediction(); assertEquals(true, Prediction.isNormPred(np)); ! } --- 29,33 ---- MixedBinaryPrediction np = new MixedBinaryPrediction(); assertEquals(true, Prediction.isNormPred(np)); ! } *************** *** 68,75 **** boolean exceptionThrown = false; try { ! p = new MixedBinaryPrediction(5); } catch (NotNormalizedPredException e) { ... [truncated message content] |
