boone.training

Class SquareError

java.lang.Object

boone.training.TrainingSignalGenerator

boone.training.SquareError

All Implemented Interfaces:

Storable, java.lang.Cloneable

public class SquareError
extends TrainingSignalGenerator

The square error function is the default for training.

(BackpropTrainer, RpropTrainer). The error term that is used is

E = 1/2 (t - o)² .

However, this function is not needed here. Instead, the Backprop trainers expect the value of a specific part of the derivative of this function. Specifically, the weight update Δw for the above error function is:

Δw = - η ⋅ ∂E/∂w, thus
Δw = - η ⋅ (t - o) ⋅ - 1 ⋅ ∂o/∂w

The back-propagation trainers only expect

error signal = (t - o)

as the so-called "error signal". The rest of the term, "- η ⋅ - 1 ⋅ ∂o/∂w", does not need to be calculated.

So for example, if another error term

E = 1/2 (t - o²)²

was used, the weight change would be

Δw = - η ⋅ ∂E/∂w = - η (t-o²) * - 2 o * ∂o/∂w

and this function would need to return

error signal = (t - o²) ⋅ 2 o .

Constructor Summary

Constructors

Constructor and Description
SquareError()

Method Summary

Modifier and Type	Method and Description
double	computeError(NeuralNet net, java.util.List<java.lang.Double> input, java.util.List<java.lang.Double> target) Computes the sum of squared errors at the output layer for the given pattern.
void	computeErrorSignal(NeuralNet net, java.util.List<java.lang.Double> input, java.util.List<java.lang.Double> target) Computes and sets the error signal for each output neuron to (output - target).

Methods inherited from class boone.training.TrainingSignalGenerator

clone, fromXML, toXML

Methods inherited from class java.lang.Object

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

Constructor Detail

SquareError

public SquareError()

Method Detail

computeErrorSignal

public void computeErrorSignal(NeuralNet net,
                               java.util.List<java.lang.Double> input,
                               java.util.List<java.lang.Double> target)

Computes and sets the error signal for each output neuron to (output - target). Note that the derivative of the neuron output is missing here, as it is later correctly multiplied to this signal (just for technical reasons).

Specified by:

computeErrorSignal in class TrainingSignalGenerator

Parameters:

net - the neural network

input - the input pattern

target - the target pattern

computeError

public double computeError(NeuralNet net,
                           java.util.List<java.lang.Double> input,
                           java.util.List<java.lang.Double> target)

Computes the sum of squared errors at the output layer for the given pattern.

Specified by:

computeError in class TrainingSignalGenerator

Parameters:

net - the neural network

input - the input pattern

target - the target pattern

Returns:

SSE