Beginning to implement neural net training. Fixed bugs in SGF/LaTeX export.

test/net/woodyfolsom/msproj/RefereeTest.java

@@ -0,0 +1,42 @@
+package net.woodyfolsom.msproj;
+
+import static org.junit.Assert.assertEquals;
+
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+
+import org.antlr.runtime.RecognitionException;
+import org.junit.Test;
+
+public class RefereeTest {
+	//private static final String FILENAME = "data/tourney1/gogame-121115115720-0500.sgf";
+	private static final String FILENAME = "data/games/1334-gokifu-20120916-Gu_Li-Lee_Sedol.sgf";
+	
+	@Test
+	public void testReplay() throws IOException, RecognitionException {
+		File sgfFile = new File(FILENAME);
+		InputStream fis = new FileInputStream(sgfFile);
+
+		GameRecord gameRecord = Referee.replay(fis);
+
+		fis.close();
+		
+		//assertEquals(9, gameRecord.getGameConfig().getSize());
+		assertEquals(19, gameRecord.getGameConfig().getSize());
+		
+		//assertEquals(5.5, gameRecord.getGameConfig().getKomi(), 0.1);
+		assertEquals(7.5, gameRecord.getGameConfig().getKomi(), 0.1);
+		
+		//assertEquals(74, gameRecord.getNumTurns());
+		assertEquals(214, gameRecord.getNumTurns());
+		
+		System.out.println("Final board state (LaTeX): ");
+		LatexWriter.write(System.out, gameRecord, 214);
+		
+		System.out.println("Final board state (SGF): ");
+		SGFWriter.write(System.out, gameRecord);
+		
+	}
+}

test/net/woodyfolsom/msproj/ann/NeuralNetLearnerTest.java

@@ -0,0 +1,86 @@
+package net.woodyfolsom.msproj.ann;
+
+import java.io.File;
+import java.util.Arrays;
+
+import org.junit.AfterClass;
+import org.junit.BeforeClass;
+import org.junit.Test;
+import org.neuroph.core.NeuralNetwork;
+import org.neuroph.core.learning.SupervisedTrainingElement;
+import org.neuroph.core.learning.TrainingSet;
+
+public class NeuralNetLearnerTest {
+	private static final String FILENAME = "myMlPerceptron.nnet";
+
+	@AfterClass
+	public static void deleteNewNet() {
+		File file = new File(FILENAME);
+		if (file.exists()) {
+			file.delete();
+		}
+	}
+
+	@BeforeClass
+	public static void deleteSavedNet() {
+		File file = new File(FILENAME);
+		if (file.exists()) {
+			file.delete();
+		}
+	}
+
+	@Test
+	public void testLearnSaveLoad() {
+		NeuralNetLearner nnLearner = new XORLearner();
+
+		// create training set (logical XOR function)
+		TrainingSet<SupervisedTrainingElement> trainingSet = new TrainingSet<SupervisedTrainingElement>(
+				2, 1);
+		for (int x = 0; x < 1000; x++) {
+		trainingSet.addElement(new SupervisedTrainingElement(new double[] { 0,
+				0 }, new double[] { 0 }));
+		trainingSet.addElement(new SupervisedTrainingElement(new double[] { 0,
+				1 }, new double[] { 1 }));
+		trainingSet.addElement(new SupervisedTrainingElement(new double[] { 1,
+				0 }, new double[] { 1 }));
+		trainingSet.addElement(new SupervisedTrainingElement(new double[] { 1,
+				1 }, new double[] { 0 }));
+		}
+		
+		nnLearner.learn(trainingSet);
+		NeuralNetwork nnet = nnLearner.getNeuralNetwork();
+
+		TrainingSet<SupervisedTrainingElement> valSet = new TrainingSet<SupervisedTrainingElement>(
+				2, 1);
+		valSet.addElement(new SupervisedTrainingElement(new double[] { 0,
+				0 }, new double[] { 0 }));
+		valSet.addElement(new SupervisedTrainingElement(new double[] { 0,
+				1 }, new double[] { 1 }));
+		valSet.addElement(new SupervisedTrainingElement(new double[] { 1,
+				0 }, new double[] { 1 }));
+		valSet.addElement(new SupervisedTrainingElement(new double[] { 1,
+				1 }, new double[] { 0 }));
+		
+		System.out.println("Output from eval set (learned network):");
+		testNetwork(nnet, valSet);
+
+		nnet.save(FILENAME);
+		nnet = NeuralNetwork.load(FILENAME);
+		
+		System.out.println("Output from eval set (learned network):");
+		testNetwork(nnet, valSet);
+	}
+	
+	private void testNetwork(NeuralNetwork nnet, TrainingSet<SupervisedTrainingElement> trainingSet) {
+		for (SupervisedTrainingElement trainingElement : trainingSet.elements()) {
+
+			nnet.setInput(trainingElement.getInput());
+			nnet.calculate();
+			double[] networkOutput = nnet.getOutput();
+			System.out.print("Input: "
+					+ Arrays.toString(trainingElement.getInput()));
+			System.out.println(" Output: " + Arrays.toString(networkOutput));
+
+		}
+	}
+}

test/net/woodyfolsom/msproj/sgf/CollectionTest.java

@@ -1,6 +1,7 @@
 package net.woodyfolsom.msproj.sgf;
 
 import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertNotNull;
 
 import java.io.ByteArrayInputStream;
 import java.io.IOException;
@@ -15,6 +16,7 @@ public class CollectionTest {
 	public static final String TEST_SGF = 
 			"(;FF[4]SZ[9]KM[5.5]RE[W+6.5]"+
 			";W[ee];B[];W[])";
+	
 	public static final String TEST_LATEX =
 			"\\white{e5}\n" +
 			"\\begin{center}\n" +
@@ -48,8 +50,6 @@ public class CollectionTest {
         CommonTokenStream tokens = new CommonTokenStream(lexer);
         SGFParser parser = new SGFParser(tokens);
         SGFNodeCollection nodeCollection = parser.collection();
-        
-        String actualLaTeX = nodeCollection.toLateX();
-        assertEquals(TEST_LATEX, actualLaTeX);
+        assertNotNull(nodeCollection);
 	}
 }

test/net/woodyfolsom/msproj/sgf/SGFParserTest.java

@@ -25,11 +25,6 @@ public class SGFParserTest {
 		System.out.println("To SGF:");
 		System.out.println(nodeCollection.toSGF());
 		System.out.println("");
-
-		System.out.println("To LaTeX:");
-		System.out.println(nodeCollection.toLateX());
-		System.out.println("");
-
 	}
 	
 	@Test
@@ -45,10 +40,5 @@ public class SGFParserTest {
 		System.out.println("To SGF:");
 		System.out.println(nodeCollection.toSGF());
 		System.out.println("");
-
-		System.out.println("To LaTeX:");
-		System.out.println(nodeCollection.toLateX());
-		System.out.println("");
-
 	}
 }