org.apache.spark.ml.classification

Class LogisticCostFun

Object

org.apache.spark.ml.classification.LogisticCostFun

All Implemented Interfaces:

breeze.linalg.ImmutableNumericOps<breeze.optimize.StochasticDiffFunction<breeze.linalg.DenseVector<Object>>>, breeze.linalg.NumericOps<breeze.optimize.StochasticDiffFunction<breeze.linalg.DenseVector<Object>>>, breeze.optimize.DiffFunction<breeze.linalg.DenseVector<Object>>, breeze.optimize.StochasticDiffFunction<breeze.linalg.DenseVector<Object>>, scala.Function1<breeze.linalg.DenseVector<Object>,Object>

public class LogisticCostFun
extends Object
implements breeze.optimize.DiffFunction<breeze.linalg.DenseVector<Object>>

LogisticCostFun implements Breeze's DiffFunction[T] for a multinomial (softmax) logistic loss function, as used in multi-class classification (it is also used in binary logistic regression). It returns the loss and gradient with L2 regularization at a particular point (coefficients). It's used in Breeze's convex optimization routines.

Nested Class Summary

Nested classes/interfaces inherited from interface breeze.linalg.NumericOps

breeze.linalg.NumericOps.Arrays$, breeze.linalg.NumericOps.ArraysLowPriority

Constructor Summary

Constructors

Constructor and Description
LogisticCostFun(RDD<org.apache.spark.ml.feature.Instance> instances, int numClasses, boolean fitIntercept, boolean standardization, Broadcast<double[]> bcFeaturesStd, double regParamL2, boolean multinomial, int aggregationDepth)

Method Summary

Modifier and Type	Method and Description
scala.Tuple2<Object,breeze.linalg.DenseVector<Object>>	calculate(breeze.linalg.DenseVector<Object> coefficients)

Methods inherited from class Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface breeze.optimize.DiffFunction

cached, repr, throughLens

Methods inherited from interface breeze.optimize.StochasticDiffFunction

apply, gradientAt, valueAt

Methods inherited from interface scala.Function1

andThen, apply$mcDD$sp, apply$mcDF$sp, apply$mcDI$sp, apply$mcDJ$sp, apply$mcFD$sp, apply$mcFF$sp, apply$mcFI$sp, apply$mcFJ$sp, apply$mcID$sp, apply$mcIF$sp, apply$mcII$sp, apply$mcIJ$sp, apply$mcJD$sp, apply$mcJF$sp, apply$mcJI$sp, apply$mcJJ$sp, apply$mcVD$sp, apply$mcVF$sp, apply$mcVI$sp, apply$mcVJ$sp, apply$mcZD$sp, apply$mcZF$sp, apply$mcZI$sp, apply$mcZJ$sp, compose, toString

Methods inherited from interface breeze.linalg.NumericOps

$amp$eq, $bar$eq, $colon$amp$eq, $colon$bar$eq, $colon$div$eq, $colon$eq, $colon$greater, $colon$greater$eq, $colon$less, $colon$less$eq, $colon$minus$eq, $colon$percent$eq, $colon$plus$eq, $colon$times$eq, $colon$up$eq, $colon$up$up$eq, $div$eq, $greater$colon$eq, $greater$colon$greater, $less$colon$eq, $less$colon$less, $minus$eq, $percent$eq, $plus, $plus$eq, $times$eq, $up$up$eq

Methods inherited from interface breeze.linalg.ImmutableNumericOps

$amp, $amp$colon$amp, $bar, $bar$colon$bar, $bslash, $colon$amp, $colon$bang$eq, $colon$bar, $colon$div, $colon$eq$eq, $colon$minus, $colon$percent, $colon$plus, $colon$times, $colon$up, $colon$up$up, $div, $div$colon$div, $minus, $minus$colon$minus, $percent, $percent$colon$percent, $plus$colon$plus, $times, $times$colon$times, $up$colon$up, $up$up, $up$up$colon$up$up, dot, t, t, t, unary_$bang, unary_$minus

Constructor Detail

LogisticCostFun

public LogisticCostFun(RDD<org.apache.spark.ml.feature.Instance> instances,
                       int numClasses,
                       boolean fitIntercept,
                       boolean standardization,
                       Broadcast<double[]> bcFeaturesStd,
                       double regParamL2,
                       boolean multinomial,
                       int aggregationDepth)

Method Detail

calculate

public scala.Tuple2<Object,breeze.linalg.DenseVector<Object>> calculate(breeze.linalg.DenseVector<Object> coefficients)

Specified by:

calculate in interface breeze.optimize.StochasticDiffFunction<breeze.linalg.DenseVector<Object>>