ODIn/spark-instrumented-optimizer
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

[SPARK-16934][ML][MLLIB] Update LogisticCostAggregator serialization code to make it consistent with LinearRegression

Browse source 
## What changes were proposed in this pull request?

Update LogisticCostAggregator serialization code to make it consistent with #14109

## How was this patch tested?
MLlib 2.0:
![image](https://cloud.githubusercontent.com/assets/19235986/17649601/5e2a79ac-61ee-11e6-833c-3bd8b5250470.png)

After this PR:
![image](https://cloud.githubusercontent.com/assets/19235986/17649599/52b002ae-61ee-11e6-9402-9feb3439880f.png)

Author: WeichenXu <WeichenXu123@outlook.com>

Closes #14520 from WeichenXu123/improve_logistic_regression_costfun.
This commit is contained in:
WeichenXu 2016-08-15 06:38:30 -07:00 committed by Yanbo Liang
parent ddf0d1e3fe
commit 3d8bfe7a39
1 changed files with 20 additions and 16 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

36
mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala
View file

@ -25,6 +25,7 @@ import org.apache.hadoop.fs.Path
import org.apache.spark.SparkException
import org.apache.spark.annotation.{Experimental, Since}
import org.apache.spark.broadcast.Broadcast
import org.apache.spark.internal.Logging
import org.apache.spark.ml.feature.Instance
import org.apache.spark.ml.linalg._
@ -346,8 +347,9 @@ class LogisticRegression @Since("1.2.0") (
val regParamL1 = $(elasticNetParam) * $(regParam)
val regParamL2 = (1.0 - $(elasticNetParam)) * $(regParam)
val bcFeaturesStd = instances.context.broadcast(featuresStd)
val costFun = new LogisticCostFun(instances, numClasses, $(fitIntercept),
$(standardization), featuresStd, featuresMean, regParamL2)
$(standardization), bcFeaturesStd, regParamL2)
val optimizer = if ($(elasticNetParam) == 0.0 || $(regParam) == 0.0) {
new BreezeLBFGS[BDV[Double]]($(maxIter), 10, $(tol))
@ -442,6 +444,7 @@ class LogisticRegression @Since("1.2.0") (
rawCoefficients(i) *= { if (featuresStd(i) != 0.0) 1.0 / featuresStd(i) else 0.0 }
i += 1
}
bcFeaturesStd.destroy(blocking = false)
if ($(fitIntercept)) {
(Vectors.dense(rawCoefficients.dropRight(1)).compressed, rawCoefficients.last,
@ -938,11 +941,15 @@ class BinaryLogisticRegressionSummary private[classification] (
* Two LogisticAggregator can be merged together to have a summary of loss and gradient of
* the corresponding joint dataset.
*
* @param bcCoefficients The broadcast coefficients corresponding to the features.
* @param bcFeaturesStd The broadcast standard deviation values of the features.
* @param numClasses the number of possible outcomes for k classes classification problem in
* Multinomial Logistic Regression.
* @param fitIntercept Whether to fit an intercept term.
*/
private class LogisticAggregator(
val bcCoefficients: Broadcast[Vector],
val bcFeaturesStd: Broadcast[Array[Double]],
private val numFeatures: Int,
numClasses: Int,
fitIntercept: Boolean) extends Serializable {
@ -958,14 +965,9 @@ private class LogisticAggregator(
* of the objective function.
*
* @param instance The instance of data point to be added.
* @param coefficients The coefficients corresponding to the features.
* @param featuresStd The standard deviation values of the features.
* @return This LogisticAggregator object.
*/
def add(
instance: Instance,
coefficients: Vector,
featuresStd: Array[Double]): this.type = {
def add(instance: Instance): this.type = {
instance match { case Instance(label, weight, features) =>
require(numFeatures == features.size, s"Dimensions mismatch when adding new instance." +
s" Expecting $numFeatures but got ${features.size}.")
@ -973,14 +975,16 @@ private class LogisticAggregator(
if (weight == 0.0) return this
val coefficientsArray = coefficients match {
val coefficientsArray = bcCoefficients.value match {
case dv: DenseVector => dv.values
case _ =>
throw new IllegalArgumentException(
s"coefficients only supports dense vector but got type ${coefficients.getClass}.")
"coefficients only supports dense vector" +
s"but got type ${bcCoefficients.value.getClass}.")
}
val localGradientSumArray = gradientSumArray
val featuresStd = bcFeaturesStd.value
numClasses match {
case 2 =>
// For Binary Logistic Regression.
@ -1077,24 +1081,23 @@ private class LogisticCostFun(
numClasses: Int,
fitIntercept: Boolean,
standardization: Boolean,
featuresStd: Array[Double],
featuresMean: Array[Double],
bcFeaturesStd: Broadcast[Array[Double]],
regParamL2: Double) extends DiffFunction[BDV[Double]] {
val featuresStd = bcFeaturesStd.value
override def calculate(coefficients: BDV[Double]): (Double, BDV[Double]) = {
val numFeatures = featuresStd.length
val coeffs = Vectors.fromBreeze(coefficients)
val bcCoeffs = instances.context.broadcast(coeffs)
val n = coeffs.size
val localFeaturesStd = featuresStd
val logisticAggregator = {
val seqOp = (c: LogisticAggregator, instance: Instance) =>
c.add(instance, coeffs, localFeaturesStd)
val seqOp = (c: LogisticAggregator, instance: Instance) => c.add(instance)
val combOp = (c1: LogisticAggregator, c2: LogisticAggregator) => c1.merge(c2)
instances.treeAggregate(
new LogisticAggregator(numFeatures, numClasses, fitIntercept)
new LogisticAggregator(bcCoeffs, bcFeaturesStd, numFeatures, numClasses, fitIntercept)
)(seqOp, combOp)
}
@ -1134,6 +1137,7 @@ private class LogisticCostFun(
}
0.5 * regParamL2 * sum
}
bcCoeffs.destroy(blocking = false)
(logisticAggregator.loss + regVal, new BDV(totalGradientArray))
}