randomize crossValidation partitions

bhklein
1 parent 89556ca3
Showing 3 changed files with 26 additions and 27 deletions
openbr/openbr_plugin.cpp
openbr/openbr_plugin.h
openbr/plugins/core/crossvalidate.cpp
@@ -529,20 +529,22 @@ QList&lt;int&gt; TemplateList::applyIndex(const QString &amp;propName, const QHash&lt;QString
     return result;
 }
-TemplateList TemplateList::partition(const QString &inputVariable, unsigned int bootStrap) const
+TemplateList TemplateList::partition(const QString &inputVariable, unsigned int randomSeed) const
 {
     TemplateList partitioned = *this;
-    if (bootStrap > 0) {
-        // Use separate RNG from Common::randN() to avoid re-seeding the global RNG.
-        // rng is seeded with the inputVariable hash in order to maintain partition across br runs, given the same bootStrap seed.
-        RandomLib::Random rng;
+    const int crossValidate = Globals->crossValidate;
+
+    // Use separate RNG from Common::randN() to avoid re-seeding the global RNG.
+    // rng is seeded with the inputVariable hash in order to maintain partition across br runs, given the same randomSeed seed.
+    RandomLib::Random rng;
+    if (crossValidate == 1) {
         for (int i=0; i < partitioned.size(); i++) {
             const QByteArray md5 = QCryptographicHash::hash(partitioned[i].file.get<QString>(inputVariable).toLatin1(), QCryptographicHash::Md5);
-            quint64 seed = md5.toHex().right(8).toULongLong(0, 16) + bootStrap;
-            rng.Reseed(seed);
-            float result = rng.FloatN();
+            quint64 labelSeed = md5.toHex().right(8).toULongLong(0, 16) + randomSeed;
+            rng.Reseed(labelSeed);
-            // Roughly 2/3rd to training, 1/3rd to testing
+            // Roughly 2/3rd to training, 1/3rd to testing for the special single split case
+            float result = rng.FloatN();
             if (result <= (2.0/3.0))
                 // Training
                 partitioned[i].file.set("Partition", QString::number(1));
@@ -550,10 +552,6 @@ TemplateList TemplateList::partition(const QString &amp;inputVariable, unsigned int 
                 partitioned[i].file.set("Partition", QString::number(0));
         }
     } else {
-        const int crossValidate = Globals->crossValidate;
-        if (crossValidate < 2)
-            return *this;
-
         for (int i=partitioned.size()-1; i>=0; i--) {
             // See CrossValidateTransform for description of these variables
             if (partitioned[i].file.getBool("duplicatePartitions")) {
@@ -565,8 +563,12 @@ TemplateList TemplateList::partition(const QString &amp;inputVariable, unsigned int 
             } else if (partitioned[i].file.getBool("allPartitions")) {
                 partitioned[i].file.set("Partition", -1);
             } else {
-                if (!partitioned[i].file.contains(("Partition"))) {
-                    const QByteArray md5 = QCryptographicHash::hash(partitioned[i].file.get<QString>(inputVariable).toLatin1(), QCryptographicHash::Md5);
+                const QByteArray md5 = QCryptographicHash::hash(partitioned[i].file.get<QString>(inputVariable).toLatin1(), QCryptographicHash::Md5);
+                if (randomSeed) {
+                    quint64 labelSeed = md5.toHex().right(8).toULongLong(0, 16) + randomSeed;
+                    rng.Reseed(labelSeed);
+                    partitioned[i].file.set("Partition", rng.Integer() % crossValidate);
+                } else if (!partitioned[i].file.contains(("Partition"))) {
                     // Select the right 8 hex characters so that it can be represented as a 64 bit integer without overflow
                     partitioned[i].file.set("Partition", md5.toHex().right(8).toULongLong(0, 16) % crossValidate);
                 }
@@ -459,7 +459,7 @@ struct TemplateList : public QList&lt;Template&gt;
     BR_EXPORT static TemplateList relabel(const TemplateList &tl, const QString &propName, bool preserveIntegers);
     /*!< \brief Assign templates to folds partitions. */
-    BR_EXPORT TemplateList partition(const QString &inputVariable, unsigned int bootStrap = 0) const;
+    BR_EXPORT TemplateList partition(const QString &inputVariable, unsigned int randomSeed = 0) const;
     BR_EXPORT QList<int> indexProperty(const QString &propName, QHash<QString, int> * valueMap=NULL,QHash<int, QVariant> * reverseLookup = NULL) const;
     BR_EXPORT QList<int> indexProperty(const QString &propName, QHash<QString, int> &valueMap, QHash<int, QVariant> &reverseLookup) const;
@@ -56,10 +56,10 @@ class CrossValidateTransform : public MetaTransform
     Q_OBJECT
     Q_PROPERTY(QString description READ get_description WRITE set_description RESET reset_description STORED false)
     Q_PROPERTY(QString inputVariable READ get_inputVariable WRITE set_inputVariable RESET reset_inputVariable STORED false)
-    Q_PROPERTY(unsigned int bootStrap READ get_bootStrap WRITE set_bootStrap RESET reset_bootStrap STORED false)
+    Q_PROPERTY(unsigned int randomSeed READ get_randomSeed WRITE set_randomSeed RESET reset_randomSeed STORED false)
     BR_PROPERTY(QString, description, "Identity")
     BR_PROPERTY(QString, inputVariable, "Label")
-    BR_PROPERTY(unsigned int, bootStrap, 0)
+    BR_PROPERTY(unsigned int, randomSeed, 0)
     // numPartitions copies of transform specified by description.
     QList<br::Transform*> transforms;
@@ -69,16 +69,12 @@ class CrossValidateTransform : public MetaTransform
     // is generally incorrect behavior.
     void train(const TemplateList &data)
     {
-        QList<int> partitions = data.partition(inputVariable, bootStrap).files().crossValidationPartitions();
-        const int numPartitions = (bootStrap > 0) ? 1 : Common::Max(partitions)+1;
+        QList<int> partitions = data.partition(inputVariable, randomSeed).files().crossValidationPartitions();
+        const int crossValidate = Globals->crossValidate;
+        const int numPartitions = (crossValidate == 1) ? 1 : Common::Max(partitions)+1;
         while (transforms.size() < numPartitions)
             transforms.append(make(description));
-        if (numPartitions < 2 && !bootStrap) {
-            transforms.first()->train(data);
-            return;
-        }
-
         QFutureSynchronizer<void> futures;
         for (int i=0; i<numPartitions; i++) {
             TemplateList partitionedData = data;
@@ -102,9 +98,10 @@ class CrossValidateTransform : public MetaTransform
     void project(const TemplateList &src, TemplateList &dst) const
     {
-        TemplateList partitioned = src.partition(inputVariable, bootStrap);
+        TemplateList partitioned = src.partition(inputVariable, randomSeed);
+        const int crossValidate = Globals->crossValidate;
-        if (bootStrap > 0) {
+        if (crossValidate == 1) {
             transforms[0]->project(partitioned, dst);
             return;
         }