Scala MultilayerPerceptronClassifier类代码示例

//设置package包名称以及导入依赖的类
package com.sircamp.algorithms.neuralnetwork

import java.io.{BufferedWriter, FileOutputStream, OutputStreamWriter}

import com.sircamp.Application
import org.apache.spark.ml.classification.{MultilayerPerceptronClassificationModel, MultilayerPerceptronClassifier}
import org.apache.spark.ml.linalg.Vector
import org.apache.spark.ml.linalg.VectorUDT
import org.apache.spark.mllib.regression.LabeledPoint
import org.apache.spark.rdd.RDD
import org.apache.spark.sql.{Dataset, Row, SparkSession}
import org.apache.spark.sql.types._


    val file = new java.io.File(TEMP_FILE_PATH)
    if( file.exists){
      file.delete()
    }

    val writer = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(file)))


    var sb = new StringBuilder()
    trainingData.collect().foreach(f=>{
      var arr = f.features.toArray
      var features = ""
      for(i <- arr.indices){
        features = features +((i+1)+":"+arr(i))+" "
      }
      writer.write((f.label+" "+features) + "\n")
    })
    writer.close()
  }

  def buildMultiPerpectronNetwork(trainingData:Dataset[Row], layers:Array[Int], maxIter:Int):MultilayerPerceptronClassificationModel = {

    val trainer = new MultilayerPerceptronClassifier()
      .setLayers(layers)
      .setBlockSize(blockSize)
      .setSeed(seed)
      .setMaxIter(maxIter)

    trainer.fit(trainingData)

  }
} 

//设置package包名称以及导入依赖的类
package cn.edu.bjtu


import org.apache.spark.SparkConf
import org.apache.spark.ml.classification.MultilayerPerceptronClassifier
import org.apache.spark.ml.evaluation.MulticlassClassificationEvaluator
import org.apache.spark.sql.SparkSession

object MLPTest {
  def main(args: Array[String]): Unit = {

    val sparkConf = new SparkConf()
      .setAppName("MLPTest")
      .setMaster("spark://master:7077")
      .setJars(Array("/home/hadoop/MLP.jar"))

    val spark = SparkSession.builder()
      .config(sparkConf)
      .getOrCreate()

    spark.sparkContext.setLogLevel("WARN")


    // Load the data stored in LIBSVM format as a DataFrame.
    val data = spark.read.format("libsvm")
      .load("hdfs://master:9000/sample_formatted.txt")

    // Split the data into train and test
    val splits = data.randomSplit(Array(0.7, 0.3), seed = 14L)
    val train = splits(0)
    val test = splits(1)

    val layers = Array[Int](20, 20, 2)

    // create the trainer and set its parameters
    val trainer = new MultilayerPerceptronClassifier()
      .setLayers(layers)
      .setBlockSize(128)
      .setSeed(14L)
      .setMaxIter(100)

    // train the model
    val model = trainer.fit(train)

    // compute accuracy on the test set
    val result = model.transform(test)
    val predictionAndLabels = result.select("prediction", "label")
    val evaluator = new MulticlassClassificationEvaluator()
      .setMetricName("accuracy")

    println("Sensitivity = " + predictionAndLabels.filter(x => x(0) == x(1) && x(0) == 1.0).count().toDouble / predictionAndLabels.filter(x => x(1) == 1.0).count().toDouble)
    println("Specificity = " + predictionAndLabels.filter(x => x(0) == x(1) && x(0) == 0.0).count().toDouble / predictionAndLabels.filter(x => x(1) == 0.0).count().toDouble)
    println("Test set accuracy = " + evaluator.evaluate(predictionAndLabels))
  }
} 

//设置package包名称以及导入依赖的类
package io.spinor.sparkdemo.mllib

import io.spinor.sparkdemo.data.MNISTData
import io.spinor.sparkdemo.util.DemoUtil
import org.apache.spark.ml.classification.MultilayerPerceptronClassifier
import org.apache.spark.ml.evaluation.MulticlassClassificationEvaluator
import org.apache.spark.ml.feature.LabeledPoint
import org.apache.spark.ml.linalg.Vectors
import org.apache.spark.sql.SparkSession
import org.apache.spark.{SparkConf, SparkContext}
import org.scalatest.{FlatSpec, Matchers}
import org.slf4j.LoggerFactory


class NeuralNetworkSpec extends FlatSpec with DemoUtil with Matchers {
  val logger = LoggerFactory.getLogger(classOf[NeuralNetworkSpec])

  "Training on MNIST data" should " run" in {
    val sparkConf = new SparkConf()
    sparkConf.setAppName("NeuralNetworkDemo")
    sparkConf.setMaster("local[2]")

    val sparkContext = new SparkContext(sparkConf)
    val sparkSession = SparkSession.builder().config(sparkConf).getOrCreate()
    val sqlContext = sparkSession.sqlContext
    import sqlContext.implicits._

    val mNISTData = new MNISTData()
    val trainingData = mNISTData.getTrainingData()
    val trainingPoints = sparkContext.parallelize(trainingData.map(entry => LabeledPoint(entry._2, Vectors.dense(entry._1)))).toDF()

    val classifier = new MultilayerPerceptronClassifier()
    classifier
      .setLayers(Array(784, 100))
      .setBlockSize(125)
      .setSeed(1234L)
      .setMaxIter(10)

    val model = classifier.fit(trainingPoints)

    val testData = mNISTData.getTestData()
    val testPoints = sparkContext.parallelize(testData.map(entry => {
    LabeledPoint(entry._2, Vectors.dense(entry._1))})).toDF()
    val result = model.transform(testPoints)
    val predictionAndLabels = result.select("prediction", "label")
    val evaluator = new MulticlassClassificationEvaluator().setMetricName("accuracy")

    logger.info("accuracy:" + evaluator.evaluate(predictionAndLabels))
  }
} 

//设置package包名称以及导入依赖的类
package com.zobot.ai.spark.helpers

import org.apache.spark.ml.classification.{MultilayerPerceptronClassificationModel, MultilayerPerceptronClassifier}
import org.apache.spark.ml.evaluation.MulticlassClassificationEvaluator
import org.apache.spark.sql.DataFrame

object MlpcHelpers {
  case class NeuralNetworkLayers (
    featureLayerSize: Int,
    intermediateLayerSizes: List[Int],
    classLayerSize: Int
  )

  def layersToArray(layers: NeuralNetworkLayers): Array[Int] = {
    (layers.featureLayerSize :: layers.intermediateLayerSizes ::: List(layers.classLayerSize)).toArray
  }

  def createTrainer(layers: NeuralNetworkLayers, blockSize: Int, maxIterations: Int, seed: Option[Long]): MultilayerPerceptronClassifier = {
    val mlpcClassifier = new MultilayerPerceptronClassifier()
      .setLayers(layersToArray(layers))
      .setBlockSize(blockSize)
      .setMaxIter(maxIterations)
    seed match {
      case Some(n) => mlpcClassifier.setSeed(n)
      case None => mlpcClassifier
    }
  }

  def trainModel(trainer: MultilayerPerceptronClassifier, trainingData: DataFrame): MultilayerPerceptronClassificationModel = {
    trainer.fit(trainingData)
  }

  def testModel(model: MultilayerPerceptronClassificationModel, testData: DataFrame): DataFrame = model.transform(testData)

  def getModelAccuracy(testResults: DataFrame): Double = {
    val predictionAndLabels = testResults.select("prediction", "label")
    val evaluator = new MulticlassClassificationEvaluator().setMetricName("accuracy")
    evaluator.evaluate(predictionAndLabels)
  }
}