org.apache.mahout.classifier.mlp
Class NeuralNetworkFunctions

java.lang.Object
  extended by org.apache.mahout.classifier.mlp.NeuralNetworkFunctions

public class NeuralNetworkFunctions
extends Object

The functions that will be used by NeuralNetwork.


Field Summary
static DoubleDoubleFunction crossEntropy
          The cross entropy function (f(t, o) = -t * log(o) - (1 - t) * log(1 - o)).
static DoubleDoubleFunction derivativeCrossEntropy
          The derivation of cross entropy function (f(t, o) = -t * log(o) - (1 - t) * log(1 - o)).
static DoubleFunction derivativeIdentityFunction
          The derivation of identity function (f(x) = x).
static DoubleDoubleFunction derivativeMinusSquared
          The derivation of minus squared function (f(t, o) = (o - t)^2).
 
Constructor Summary
NeuralNetworkFunctions()
           
 
Method Summary
static DoubleDoubleFunction getDerivativeDoubleDoubleFunction(String function)
          Get the corresponding derivation of double double function by the name.
static DoubleFunction getDerivativeDoubleFunction(String function)
          Get the derivation double function by the name.
static DoubleDoubleFunction getDoubleDoubleFunction(String function)
          Get the corresponding double-double function by the name.
static DoubleFunction getDoubleFunction(String function)
          Get the corresponding function by its name.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

derivativeIdentityFunction

public static DoubleFunction derivativeIdentityFunction
The derivation of identity function (f(x) = x).


derivativeMinusSquared

public static DoubleDoubleFunction derivativeMinusSquared
The derivation of minus squared function (f(t, o) = (o - t)^2).


crossEntropy

public static DoubleDoubleFunction crossEntropy
The cross entropy function (f(t, o) = -t * log(o) - (1 - t) * log(1 - o)).


derivativeCrossEntropy

public static DoubleDoubleFunction derivativeCrossEntropy
The derivation of cross entropy function (f(t, o) = -t * log(o) - (1 - t) * log(1 - o)).

Constructor Detail

NeuralNetworkFunctions

public NeuralNetworkFunctions()
Method Detail

getDoubleFunction

public static DoubleFunction getDoubleFunction(String function)
Get the corresponding function by its name. Currently supports: "Identity", "Sigmoid".

Parameters:
function - The name of the function.
Returns:
The corresponding double function.

getDerivativeDoubleFunction

public static DoubleFunction getDerivativeDoubleFunction(String function)
Get the derivation double function by the name. Currently supports: "Identity", "Sigmoid".

Parameters:
function - The name of the function.
Returns:
The double function.

getDoubleDoubleFunction

public static DoubleDoubleFunction getDoubleDoubleFunction(String function)
Get the corresponding double-double function by the name. Currently supports: "Minus_Squared", "Cross_Entropy".

Parameters:
function - The name of the function.
Returns:
The double-double function.

getDerivativeDoubleDoubleFunction

public static DoubleDoubleFunction getDerivativeDoubleDoubleFunction(String function)
Get the corresponding derivation of double double function by the name. Currently supports: "Minus_Squared", "Cross_Entropy".

Parameters:
function - The name of the function.
Returns:
The double-double-function.


Copyright © 2008–2014 The Apache Software Foundation. All rights reserved.