DL4J Scala Example

Iris.scala

import org.deeplearning4j.nn.conf.NeuralNetConfiguration;
import org.deeplearning4j.nn.conf.layers.RBM;
import org.deeplearning4j.nn.weights.WeightInit;
import org.deeplearning4j.nn.conf.distribution.UniformDistribution;
import org.deeplearning4j.nn.api.OptimizationAlgorithm;
import org.nd4j.linalg.lossfunctions.LossFunctions;
import org.deeplearning4j.nn.conf.`override`.ClassifierOverride;
import org.deeplearning4j.nn.multilayer.MultiLayerNetwork;
import org.deeplearning4j.optimize.listeners.ScoreIterationListener;
import org.deeplearning4j.optimize.api.IterationListener;
import scala.collection.JavaConversions._
import org.nd4j.linalg.factory.Nd4j;
import org.deeplearning4j.datasets.iterator.impl.IrisDataSetIterator;
import org.deeplearning4j.eval.Evaluation;


object Main extends App {
  Nd4j.MAX_SLICES_TO_PRINT = -1
  Nd4j.MAX_ELEMENTS_PER_SLICE = -1

  val conf = new NeuralNetConfiguration.Builder()
    .iterations(100).layer(new RBM())
    .weightInit(WeightInit.DISTRIBUTION).dist(new UniformDistribution(0, 1))
    .activationFunction("tanh").momentum(0.9)
    .optimizationAlgo(OptimizationAlgorithm.LBFGS)
    .constrainGradientToUnitNorm(true).k(1).regularization(true).l2(2e-4)
    .visibleUnit(RBM.VisibleUnit.GAUSSIAN).hiddenUnit(RBM.HiddenUnit.RECTIFIED)
    .lossFunction(LossFunctions.LossFunction.RMSE_XENT)
    .learningRate(1e-1f)
    .nIn(4).nOut(3).list(2)
    .hiddenLayerSizes(3)
    .`override`(new ClassifierOverride(1)).build()


  val d = new MultiLayerNetwork(conf)
  d.init()
  d.setListeners(seqAsJavaList(List((new ScoreIterationListener(1)).asInstanceOf[IterationListener])))

  val iter = new IrisDataSetIterator(150, 150)

  val next = iter.next();

  Nd4j.writeTxt(next.getFeatureMatrix(), "iris.txt", "\t");

  next.normalizeZeroMeanZeroUnitVariance();

  val testAndTrain = next.splitTestAndTrain(110);
  val train = testAndTrain.getTrain();

  d.fit(train);

  val test = testAndTrain.getTest()
  val eval = new Evaluation()
  val output = d.output(test.getFeatureMatrix())
  eval.eval(test.getLabels(), output)

  println(s"Score ${eval.stats}")

}

This works with deeplearning4j-core version: "0.0.3.3.4.alpha1-SNAPSHOT", so you may need to clone the repo local and run mvn install -Dmaven.test.skip=true first.

Then the build.sbt should look something like this:

val dl4jVersion = "0.0.3.5.5.3"

val deeplearnVersion = "0.0.3.3.4.alpha1-SNAPSHOT"

resolvers += "Local Maven Repo" at "file:///Users/dosht/.m2/repository/"

libraryDependencies ++= Seq(
  "org.nd4j" % "nd4j-jblas" % dl4jVersion,
  "org.deeplearning4j" % "deeplearning4j-core" % deeplearnVersion
)