org.apache.mahout.classifier.sgd

Class L1

java.lang.Object

org.apache.mahout.classifier.sgd.L1

All Implemented Interfaces:

org.apache.hadoop.io.Writable, PriorFunction

public class L1
extends Object
implements PriorFunction

Implements the Laplacian or bi-exponential prior. This prior has a strong tendency to set coefficients to zero and thus is useful as an alternative to variable selection. This version implements truncation which prevents a coefficient from changing sign. If a correction would change the sign, the coefficient is truncated to zero. Note that it doesn't matter to have a scale for this distribution because after taking the derivative of the logP, the lambda coefficient used to combine the prior with the observations has the same effect. If we had a scale here, then it would be the same effect as just changing lambda.

Constructor Summary

Constructors

Constructor and Description
L1()

Method Summary

Modifier and Type	Method and Description
double	age(double oldValue, double generations, double learningRate) Applies the regularization to a coefficient.
double	logP(double betaIJ) Returns the log of the probability of a particular coefficient value according to the prior.
void	readFields(DataInput dataInput)
void	write(DataOutput out)

Methods inherited from class java.lang.Object

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

Constructor Detail

L1

public L1()

Method Detail

age

public double age(double oldValue,
                  double generations,
                  double learningRate)

Description copied from interface: PriorFunction

Applies the regularization to a coefficient.

Specified by:

age in interface PriorFunction

Parameters:

oldValue - The previous value.

generations - The number of generations.

learningRate - The learning rate with lambda baked in.

Returns:

The new coefficient value after regularization.

logP

public double logP(double betaIJ)

Description copied from interface: PriorFunction

Returns the log of the probability of a particular coefficient value according to the prior.

Specified by:

logP in interface PriorFunction

Parameters:

betaIJ - The coefficient.

Returns:

The log probability.

write

public void write(DataOutput out)
           throws IOException

Specified by:

write in interface org.apache.hadoop.io.Writable

Throws:

IOException

readFields

public void readFields(DataInput dataInput)
                throws IOException

Specified by:

readFields in interface org.apache.hadoop.io.Writable

Throws:

IOException