Class Updater

Object
org.apache.spark.mllib.optimization.Updater
All Implemented Interfaces:
Serializable
Direct Known Subclasses:
L1Updater, SimpleUpdater, SquaredL2Updater

public abstract class Updater extends Object implements Serializable
Class used to perform steps (weight update) using Gradient Descent methods.

For general minimization problems, or for regularized problems of the form min L(w) + regParam * R(w), the compute function performs the actual update step, when given some (e.g. stochastic) gradient direction for the loss L(w), and a desired step-size (learning rate).

The updater is responsible to also perform the update coming from the regularization term R(w) (if any regularization is used).

See Also:
  • Constructor Summary

    Constructors
    Constructor
    Description
     
  • Method Summary

    Modifier and Type
    Method
    Description
    abstract scala.Tuple2<Vector,Object>
    compute(Vector weightsOld, Vector gradient, double stepSize, int iter, double regParam)
    Compute an updated value for weights given the gradient, stepSize, iteration number and regularization parameter.

    Methods inherited from class java.lang.Object

    equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
  • Constructor Details

    • Updater

      public Updater()
  • Method Details

    • compute

      public abstract scala.Tuple2<Vector,Object> compute(Vector weightsOld, Vector gradient, double stepSize, int iter, double regParam)
      Compute an updated value for weights given the gradient, stepSize, iteration number and regularization parameter. Also returns the regularization value regParam * R(w) computed using the *updated* weights.

      Parameters:
      weightsOld - - Column matrix of size dx1 where d is the number of features.
      gradient - - Column matrix of size dx1 where d is the number of features.
      stepSize - - step size across iterations
      iter - - Iteration number
      regParam - - Regularization parameter

      Returns:
      A tuple of 2 elements. The first element is a column matrix containing updated weights, and the second element is the regularization value computed using updated weights.