Integrate semi-supervised learning to MOA

dumpClustering.csv

@@ -0,0 +1 @@
+Nr;Event;

expe/cl-categorical.sh

@@ -0,0 +1,64 @@
+########################################################
+#                                                      #
+# CLUSTER-AND-LABEL WITH DIFFERENT CATEGORICAL METRICS #
+#                                                      # 
+########################################################
+
+rootfolder=/mnt/d/Study/M2_DK/Internship
+folder="cl-categorical"
+m=100
+h=1000
+evaluator="BasicClassificationPerformanceEvaluator -o -p -r -f"
+learner_sp="trees.HoeffdingTree"
+global_learner="trees.HoeffdingTree"
+local_learner="bayes.NaiveBayes"
+
+for d in randomTree elecNormNew airlines;
+# for d in agrawal;
+do
+	for ratio in 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99;
+	do
+		# compute the percentage of unlabeled data
+		frac=`echo $ratio | cut -d. -f2`
+		len=${#frac}
+		if [[ $len == 1 ]]; then
+			((percent=frac*10))
+		else
+			percent=$frac
+		fi
+
+		# define necessary variables
+		data="$rootfolder/data/semi-masked/$d-semi-$percent.arff"
+		stream="ArffFileStream -f $data"
+
+		# define the task
+		for method in 0 1 2 3 4 5;
+		do
+			echo
+			echo "==============================================================================================="
+			echo "CLUSTER-N-LABEL & MULTIVIEW & CATEGORICAL METRICS: dataset = $d, r = $percent, method = $method"
+			echo "==============================================================================================="
+			echo
+
+			if [[ $method == 0 ]]; then
+				method_name="nothing"
+			elif [[ $method == 1 ]]; then
+				method_name="euclidean"
+			elif [[ $method == 2 ]]; then
+				method_name="of"
+			elif [[ $method == 3 ]]; then
+				method_name="lin"
+			elif [[ $method == 4 ]]; then
+				method_name="goodall3"
+			elif [[ $method == 5 ]]; then
+				method_name="iof"
+			fi
+
+			clusterer="semisupervised.ClustreamSSL -h $h -l 0.0 -m $m -k 5 -a $method"
+			learner_ssl="semisupervised.ClusterAndLabelSubLearnerClassifier -c ($clusterer) -p -l ($local_learner) -g ($global_learner) -t 1"
+			result="$rootfolder/experiments/$folder/$d-cl-$percent-$method_name.csv"
+			task="EvaluateInterleavedTestThenTrainSemi -b ($learner_sp) -l ($learner_ssl) -s ($stream) -d ($result) -i -1 -f 1000 -q 1000"
+			java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+		done
+	done
+done

expe/cl-ensemble.sh

@@ -0,0 +1,45 @@
+###################################
+#                                 #
+# CLUSTER-AND-LABEL WITH ENSEMBLE #
+#                                 #
+###################################
+
+rootfolder=/mnt/d/Study/M2_DK/Internship
+folder="cl-ensemble"
+m=100
+h=1000
+k=1
+evaluator="BasicClassificationPerformanceEvaluator -o -p -r -f"
+learner_sp="trees.HoeffdingTree"
+clusterer="semisupervised.ClustreamSSL -h $h -l 0.0 -m $m -k 5 -a 1"
+learner_ssl="semisupervised.ClusterAndLabelClassifier -c ($clusterer) -p -k 1"
+
+# for d in led randomRBF randomTree elecNormNew covtypeNorm airlines;
+for d in agrawal hyperplane;
+do
+	for ratio in 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99;
+	do
+		# compute the percentage of unlabeled data
+		frac=`echo $ratio | cut -d. -f2`
+		len=${#frac}
+		if [[ $len == 1 ]]; then
+			((percent=frac*10))
+		else
+			percent=$frac
+		fi
+
+		# define necessary variables
+		data="$rootfolder/data/semi-masked/$d-semi-$percent.arff"
+		stream="ArffFileStream -f $data"
+
+		echo
+		echo "=================================================================================="
+		echo "CLUSTER-N-LABEL, ENSEMBLE WITH LEVERAGE BAGGING: dataset = $d, r = $percent"
+		echo "=================================================================================="
+		echo
+		ensemble="meta.LeveragingBag -l ($learner_ssl)"
+		result="$rootfolder/experiments/$folder/$d-cl-$percent-$m-$h-$k-leveragebag.csv"
+		task="EvaluateInterleavedTestThenTrain -l ($ensemble) -s ($stream) -d ($result) -e ($evaluator) -i -1 -f 1000 -q 1000"
+		java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+	done
+done

expe/cl-sublearner-forcepred.sh

@@ -0,0 +1,45 @@
+###################################################################
+#                                   															#
+# CLUSTER-AND-LABEL WITH SUBLEARNER AND FORCED CORRECT PREDICTION #
+#                                   															# 
+###################################################################
+
+rootfolder=/mnt/d/Study/M2_DK/Internship
+folder="cl-sublearner"
+m=100
+h=1000
+
+for d in agrawal led randomTree randomRBF hyperplane elecNormNew covtypeNorm airlines;
+do
+	for ratio in 0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99;
+	do
+		# compute the percentage of unlabeled data
+		frac=`echo $ratio | cut -d. -f2`
+		len=${#frac}
+		if [[ $len == 1 ]]; then
+			((percent=frac*10))
+		else
+			percent=$frac
+		fi
+
+		# define necessary variables
+		data="$rootfolder/data/semi-masked/$d-semi-$percent.arff"
+		evaluator="BasicClassificationPerformanceEvaluator -o -p -r -f"
+		learner_sp="trees.HoeffdingTree"
+		clusterer="semisupervised.ClustreamSSL -h $h -l 0.0 -m $m -k 5 -a 1" # still use Euclidean distance
+		global_learner="trees.HoeffdingTree"
+		local_learner="bayes.NaiveBayes"
+		stream="ArffFileStream -f $data"
+
+		# define the task
+		echo
+		echo "=================================================================================="
+		echo "CLUSTER-N-LABEL & HETERO & FORCE PREDICTION: dataset = $d, r = $percent"
+		echo "=================================================================================="
+		echo
+		learner_ssl="semisupervised.ClusterAndLabelSubLearnerClassifier -c ($clusterer) -p -l ($local_learner) -g ($global_learner) -t 2"
+		result="$rootfolder/experiments/$folder/$d-cl-$percent-hetero-forced-prediction.csv"
+		task="EvaluateInterleavedTestThenTrainSemi -b ($learner_sp) -l ($learner_ssl) -s ($stream) -d ($result) -e ($evaluator) -i -1 -f 1000 -q 1000"
+		java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+	done
+done

expe/cl-sublearner-measures.sh

@@ -0,0 +1,47 @@
+################################################################
+#                                                              #
+# CLUSTER-AND-LABEL + HETEROGENEOUS LEARNERS + MORE STATISTICS #
+#                                                              # 
+################################################################
+
+rootfolder=/mnt/d/Study/M2_DK/Internship
+folder="cl-sublearner"
+m=100
+h=1000
+evaluator="BasicClassificationPerformanceEvaluator -o -p -r -f"
+learner_sp="trees.HoeffdingTree"
+clusterer="semisupervised.ClustreamSSL -h $h -l 0.0 -m $m -k 5 -a 1"
+global_learner="trees.HoeffdingTree"
+local_learner="bayes.NaiveBayes"
+
+for d in agrawal led randomTree randomRBF hyperplane elecNormNew covtypeNorm airlines;
+# for d in elecNormNew;
+do
+	#for ratio in 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99;
+	for ratio in 0.90 0.95 0.99;
+	do
+		# compute the percentage of unlabeled data
+		frac=`echo $ratio | cut -d. -f2`
+		len=${#frac}
+		if [[ $len == 1 ]]; then
+			((percent=frac*10))
+		else
+			percent=$frac
+		fi
+
+		# define necessary variables
+		data="$rootfolder/data/semi-masked/$d-semi-$percent.arff"
+		stream="ArffFileStream -f $data"
+
+		# define the task
+		echo
+		echo "=================================================================================="
+		echo "CLUSTER-N-LABEL & HETEROGENEOUS LEARNERS: dataset = $d, r = $percent"
+		echo "=================================================================================="
+		echo
+		learner_ssl="semisupervised.ClusterAndLabelSubLearnerClassifier -c ($clusterer) -p -l ($local_learner) -g ($global_learner) -t 1"
+		result="$rootfolder/experiments/$folder/$d-cl-$percent-hetero-measures-euclidean.csv"
+		task="EvaluateInterleavedTestThenTrainSemi -b ($learner_sp) -l ($learner_ssl) -s ($stream) -d ($result) -e ($evaluator) -i -1 -f 1000 -q 1000"
+		java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+	done
+done

expe/cl-sublearner.sh

@@ -0,0 +1,65 @@
+#####################################
+#                                   #
+# CLUSTER-AND-LABEL WITH SUBLEARNER #
+#                                   # 
+#####################################
+
+rootfolder=/mnt/d/Study/M2_DK/Internship
+folder="cl-sublearner"
+m=100
+h=1000
+
+for d in agrawal led randomTree randomRBF hyperplane elecNormNew covtypeNorm airlines;
+do
+	for ratio in 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99;
+	do
+		# compute the percentage of unlabeled data
+		frac=`echo $ratio | cut -d. -f2`
+		len=${#frac}
+		if [[ $len == 1 ]]; then
+			((percent=frac*10))
+		else
+			percent=$frac
+		fi
+
+		# define necessary variables
+		data="$rootfolder/data/semi-masked/$d-semi-$percent.arff"
+		evaluator="BasicClassificationPerformanceEvaluator -o -p -r -f"
+		learner_sp="trees.HoeffdingTree"
+		clusterer="semisupervised.ClustreamSSL -h $h -l 0.0 -m $m -k 5 -a 1"
+		global_learner="trees.HoeffdingTree"
+		local_learner="bayes.NaiveBayes"
+		stream="ArffFileStream -f $data"
+
+		# define the task
+		echo
+		echo "=================================================================================="
+		echo "CLUSTER-N-LABEL & MULTIVIEW: dataset = $d, r = $percent"
+		echo "=================================================================================="
+		echo
+		learner_ssl="semisupervised.ClusterAndLabelSubLearnerClassifier -c ($clusterer) -p -l ($local_learner) -g ($global_learner) -t 1"
+		result="$rootfolder/experiments/$folder/$d-cl-$percent-multiview.csv"
+		task="EvaluateInterleavedTestThenTrainSemi -b ($learner_sp) -l ($learner_ssl) -s ($stream) -d ($result) -i -1 -f 1000 -q 1000"
+		java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+
+		echo
+		echo "=================================================================================="
+		echo "CLUSTER-N-LABEL STANDARD: dataset = $d, r = $percent"
+		echo "=================================================================================="
+		echo
+		learner_ssl="semisupervised.ClusterAndLabelClassifier -c ($clusterer) -p -k 1"
+		result="$rootfolder/experiments/$folder/$d-cl-$percent-standard.csv"
+		task="EvaluateInterleavedTestThenTrainSemi -b ($learner_sp) -l ($learner_ssl) -s ($stream) -d ($result) -i -1 -f 1000 -q 1000"
+		java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+
+		echo
+		echo "=================================================================================="
+		echo "FULLY SUPERVISED: dataset = $d, r = $percent"
+		echo "=================================================================================="
+		echo
+		learner_ssl="semisupervised.ClusterAndLabelSubLearnerClassifier -c ($clusterer) -p -l ($local_learner) -g ($global_learner) -t 1"
+		result="$rootfolder/experiments/$folder/$d-cl-$percent-supervised.csv"
+		task="EvaluateInterleavedTestThenTrainSemi -a -b ($learner_sp) -l ($learner_ssl) -s ($stream) -d ($result) -i -1 -f 1000 -q 1000"
+		java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask $task
+	done
+done

expe/test.sh

@@ -0,0 +1 @@
+java -classpath /mnt/d/Study/M2_DK/Internship/work/moa/moa/target/classes moa.DoTask

moa/.classpath

@@ -24,9 +24,8 @@
 			<attribute name="maven.pomderived" value="true"/>
 		</attributes>
 	</classpathentry>
-	<classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-10">
+	<classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.8">
 		<attributes>
-			<attribute name="module" value="true"/>
 			<attribute name="maven.pomderived" value="true"/>
 		</attributes>
 	</classpathentry>

moa/.settings/org.eclipse.jdt.core.prefs

@@ -1,13 +1,13 @@
 eclipse.preferences.version=1
 org.eclipse.jdt.core.compiler.codegen.inlineJsrBytecode=enabled
-org.eclipse.jdt.core.compiler.codegen.targetPlatform=10
+org.eclipse.jdt.core.compiler.codegen.targetPlatform=1.8
 org.eclipse.jdt.core.compiler.codegen.unusedLocal=preserve
-org.eclipse.jdt.core.compiler.compliance=10
+org.eclipse.jdt.core.compiler.compliance=1.8
 org.eclipse.jdt.core.compiler.debug.lineNumber=generate
 org.eclipse.jdt.core.compiler.debug.localVariable=generate
 org.eclipse.jdt.core.compiler.debug.sourceFile=generate
 org.eclipse.jdt.core.compiler.problem.assertIdentifier=error
 org.eclipse.jdt.core.compiler.problem.enumIdentifier=error
 org.eclipse.jdt.core.compiler.problem.forbiddenReference=warning
 org.eclipse.jdt.core.compiler.release=enabled
-org.eclipse.jdt.core.compiler.source=10
+org.eclipse.jdt.core.compiler.source=1.8