dl4j_scala_jupyter

%AddJar http://central.maven.org/maven2/org/deeplearning4j/deeplearning4j-core/0.4.0/deeplearning4j-core-0.4.0.jar
%AddJar http://central.maven.org/maven2/org/nd4j/nd4j-api/0.4.0/nd4j-api-0.4.0.jar
%AddJar http://central.maven.org/maven2/org/nd4j/nd4j-buffer/0.4.0/nd4j-buffer-0.4.0.jar
%AddJar http://central.maven.org/maven2/org/nd4j/canova-nd4j-image/0.0.0.17/canova-nd4j-image-0.0.0.17.jar
%AddJar http://central.maven.org/maven2/org/nd4j/canova-nd4j-codec/0.0.0.17/canova-nd4j-codec-0.0.0.17.jar
%AddJar http://central.maven.org/maven2/org/nd4j/canova-api/0.0.0.17/canova-api-0.0.0.17.jar
%AddJar http://central.maven.org/maven2/commons-io/commons-io/2.4/commons-io-2.4.jar
%AddJar http://central.maven.org/maven2/org/nd4j/nd4j-x86/0.4-rc3.8/nd4j-x86-0.4-rc3.8.jar

import java.io.File;
import org.apache.commons.io.FileUtils
import java.net.URL
import org.canova.api.records.reader.RecordReader;
import org.canova.api.records.reader.impl.CSVRecordReader;
import org.canova.api.split.FileSplit;
import org.deeplearning4j.datasets.canova.RecordReaderDataSetIterator;
import org.deeplearning4j.eval.Evaluation;
import org.deeplearning4j.nn.api.OptimizationAlgorithm;
import org.deeplearning4j.nn.conf.MultiLayerConfiguration;
import org.deeplearning4j.nn.conf.NeuralNetConfiguration;
import org.deeplearning4j.nn.conf.Updater;
import org.deeplearning4j.nn.conf.layers.DenseLayer;
import org.deeplearning4j.nn.conf.layers.OutputLayer;
import org.deeplearning4j.nn.multilayer.MultiLayerNetwork;
import org.deeplearning4j.nn.weights.WeightInit;
import org.deeplearning4j.optimize.listeners.ScoreIterationListener;
import org.nd4j.linalg.api.ndarray.INDArray;
import org.nd4j.linalg.dataset.DataSet;
import org.nd4j.linalg.dataset.api.iterator.DataSetIterator;
import org.nd4j.linalg.lossfunctions.LossFunctions.LossFunction;

val seed = 123;
val learningRate = 0.01;
val batchSize = 1000;
val nEpochs = 30;

val numInputs = 3;
val numOutputs = 7;
val numHiddenNodes = 3;

val rr = new CSVRecordReader();

val url = "http://github.com/romeokienzler/developerWorks/raw/master/data_dl4j.csv"
val tempDir = System.getProperty("java.io.tmpdir")
val fileLocation = tempDir + "data_dl4j.csv"
val f = new File(fileLocation)
if (!f.exists()) {
    FileUtils.copyURLToFile(new URL(url), f)
    println("File downloaded to " + f.getAbsolutePath());
} else {
    println("Using existing text file at " + f.getAbsolutePath());
}

rr.initialize(new FileSplit(new File(fileLocation)));


val trainIter = new RecordReaderDataSetIterator(rr,batchSize, 0, 7);

val rrTest = new CSVRecordReader();
rrTest.initialize(new FileSplit(new File(fileLocation)));
val testIter = new RecordReaderDataSetIterator(rrTest,batchSize, 0, 7);


val conf=new NeuralNetConfiguration.Builder().seed(seed).iterations(1).optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).learningRate(learningRate).updater(Updater.NESTEROVS).momentum(0.9).list().layer(0,new DenseLayer.Builder().nIn(numInputs).nOut(numHiddenNodes).weightInit(WeightInit.XAVIER).activation("relu").build()).layer(1,new DenseLayer.Builder().nIn(numHiddenNodes).nOut(numHiddenNodes).weightInit(WeightInit.XAVIER).activation("sigmoid").build()).layer(2,new DenseLayer.Builder().nIn(numHiddenNodes).nOut(numHiddenNodes).weightInit(WeightInit.XAVIER).activation("sigmoid").build()).layer(3,new OutputLayer.Builder(LossFunction.NEGATIVELOGLIKELIHOOD).weightInit(WeightInit.XAVIER).activation("softmax").weightInit(WeightInit.XAVIER).nIn(numHiddenNodes).nOut(numOutputs).build()).pretrain(false).backprop(true).build();