flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-1807) Stochastic gradient descent optimizer for ML library
Date Thu, 07 May 2015 09:39:00 GMT

    [ https://issues.apache.org/jira/browse/FLINK-1807?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14532336#comment-14532336
] 

ASF GitHub Bot commented on FLINK-1807:
---------------------------------------

Github user tillrohrmann commented on a diff in the pull request:

    https://github.com/apache/flink/pull/613#discussion_r29837353
  
    --- Diff: flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/optimization/RegularizationType.scala
---
    @@ -0,0 +1,171 @@
    +/*
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + *     http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +package org.apache.flink.ml.optimization
    +
    +import org.apache.flink.api.scala._
    +import org.apache.flink.ml.math.{Vector => FlinkVector, BLAS}
    +import org.apache.flink.ml.math.Breeze._
    +
    +import breeze.numerics._
    +import breeze.linalg.{norm => BreezeNorm, max => BreezeMax}
    +
    +
    +
    +// TODO(tvas): Change name to RegularizationPenalty?
    +/** Represents a type of regularization penalty
    +  *
    +  * Regularization penalties are used to restrict the optimization problem to solutions
with
    +  * certain desirable characteristics, such as sparsity for the L1 penalty, or penalizing
large
    +  * weights for the L2 penalty.
    +  *
    +  * The regularization term, $R(w)$ is added to the objective function, $f(w) = L(w)
+ \lambda R(w)$
    +  * where $\lambda$ is the regularization parameter used to tune the amount of regularization
    +  * applied.
    +  */
    +abstract class RegularizationType extends Serializable {
    +
    +  /** Updates the weights by taking a step according to the gradient and regularization
applied
    +    *
    +    * @param oldWeights The weights to be updated
    +    * @param gradient The gradient according to which we will update the weights
    +    * @param effectiveStepSize The effective step size for this iteration
    +    * @param regParameter The regularization parameter to be applied in the case of L1
    +    *                     regularization
    +    */
    +  def takeStep(
    +      oldWeights: FlinkVector,
    +      gradient: FlinkVector,
    +      effectiveStepSize: Double,
    +      regParameter: Double) {
    +    BLAS.axpy(-effectiveStepSize, gradient, oldWeights)
    +  }
    +
    +  /** Adds regularization to the loss value **/
    +  def regLoss(oldLoss: Double, weightVector: FlinkVector, regularizationParameter: Double):
Double
    +
    +}
    +
    +/** Abstract class for regularization penalties that are differentiable
    +  *
    +  */
    +abstract class DiffRegularizationType extends RegularizationType {
    +
    +  /** Compute the regularized gradient loss for the given data.
    +    * The provided cumGradient is updated in place.
    +    *
    +    * @param weightVector The current weight vector
    +    * @param lossGradient The vector to which the gradient will be added to, in place.
    +    * @return The regularized loss. The gradient is updated in place.
    +    */
    +  def regularizedLossAndGradient(
    +      loss: Double,
    +      weightVector: FlinkVector,
    +      lossGradient: FlinkVector,
    +      regularizationParameter: Double) : Double ={
    +    val adjustedLoss = regLoss(loss, weightVector, regularizationParameter)
    +    regGradient(weightVector, lossGradient, regularizationParameter)
    +
    +    adjustedLoss
    +  }
    +
    +  /** Adds regularization gradient to the loss gradient. The gradient is updated in place
**/
    +  def regGradient(
    +      weightVector: FlinkVector,
    +      lossGradient: FlinkVector,
    +      regularizationParameter: Double)
    +}
    +
    +/** Performs no regularization, equivalent to $R(w) = 0$ **/
    +class NoRegularization extends RegularizationType {
    +  /** Adds regularization to the loss value **/
    +  override def regLoss(oldLoss: Double, weightVector: FlinkVector, regularizationParameter:
Double):
    +  Double = {oldLoss}
    +}
    +
    +/** $L_2$ regularization penalty.
    +  *
    +  * Penalizes large weights, favoring solutions with more small weights rather than few
large ones.
    +  *
    +  */
    +class L2Regularization extends DiffRegularizationType {
    +
    +  /** Adds regularization to the loss value **/
    +  override def regLoss(oldLoss: Double, weightVector: FlinkVector, regParameter: Double)
    +    : Double = {
    +    val brzVector = weightVector.asBreeze
    +    oldLoss + regParameter * (brzVector dot brzVector) / 2
    +  }
    +
    +  /** Adds regularization gradient to the loss gradient. The gradient is updated in place
**/
    +  override def regGradient(
    +      weightVector: FlinkVector,
    +      lossGradient: FlinkVector,
    +      regParameter: Double): Unit = {
    +    BLAS.axpy(regParameter, weightVector, lossGradient)
    +  }
    +}
    +
    +/** $L_1$ regularization penalty.
    +  *
    +  * The $L_1$ penalty can be used to drive a number of the solution coefficients to 0,
thereby
    +  * producing sparse solutions.
    +  *
    +  */
    +class L1Regularization extends RegularizationType {
    +  /** Calculates and applies the regularization amount and the regularization parameter
    +    *
    +    * Implementation was taken from the Apache Spark Mllib library:
    +    * http://git.io/vfZIT
    +    * @param oldWeights The old weights
    +    * @param effectiveStepSize The effective step size for this iteration
    +    * @param regParameter The current regularization parameter
    +    * @return A tuple whose first element is the updated weight FlinkVector and the second
is the
    +    *         regularization value
    +    */
    +  override def takeStep(
    +      oldWeights: FlinkVector,
    +      gradient: FlinkVector,
    +      effectiveStepSize: Double,
    +      regParameter: Double) {
    +    BLAS.axpy(-effectiveStepSize, gradient, oldWeights)
    +    val brzWeights = oldWeights.asBreeze.toDenseVector
    +
    +    // Apply proximal operator (soft thresholding)
    +    val shrinkageVal = regParameter * effectiveStepSize
    +    var i = 0
    +    while (i < brzWeights.length) {
    +      val wi = brzWeights(i)
    +      brzWeights(i) = signum(wi) * BreezeMax(0.0, abs(wi) - shrinkageVal)
    +      i += 1
    +    }
    +
    +    BLAS.copy(brzWeights.fromBreeze, oldWeights)
    --- End diff --
    
    Why using for some operations Breeze and for others the BLAS object? With Breeze you should
be able to do at least the same as with the BLAS object, right? Wouldn't it make sense to
do all with either Breeze or BLAS? That way we would only have one "dependency".


> Stochastic gradient descent optimizer for ML library
> ----------------------------------------------------
>
>                 Key: FLINK-1807
>                 URL: https://issues.apache.org/jira/browse/FLINK-1807
>             Project: Flink
>          Issue Type: Improvement
>          Components: Machine Learning Library
>            Reporter: Till Rohrmann
>            Assignee: Theodore Vasiloudis
>              Labels: ML
>
> Stochastic gradient descent (SGD) is a widely used optimization technique in different
ML algorithms. Thus, it would be helpful to provide a generalized SGD implementation which
can be instantiated with the respective gradient computation. Such a building block would
make the development of future algorithms easier.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message