com.penguinwerks.jodene.activation
Class SigmoidActivationFunction

java.lang.Object
  com.penguinwerks.jodene.activation.SigmoidActivationFunction

All Implemented Interfaces:

ActivationFunction, java.io.Serializable

public class SigmoidActivationFunction

extends java.lang.Object

implements ActivationFunction, java.io.Serializable

The sigmoid activation function. This is probably the most popular non-linear squashing function used in neural networks. The activation value is computed as per the sigmoid function 1/(1 + exp(-x)). The output is then scaled between the min and max values.

Author:

Paul Hoehne

See Also:

Serialized Form

Constructor Summary

SigmoidActivationFunction()
Default constructor.

Method Summary

double	activation(double val) Calculates the activation value as per the sigmoid function.
double	derivative(double val) Calculate the derivative of the activation value.
double	getMax() The maximum value for activation.
double	getMin() The minimum value for activation.
static void	main(java.lang.String[] args)
void	setMax(double val) The maximum value for output activation.
void	setMin(double val) The mimimum output value for activation.

Methods inherited from class java.lang.Object

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

Constructor Detail

SigmoidActivationFunction

public SigmoidActivationFunction()

Default constructor.

Method Detail

setMin

public void setMin(double val)

The mimimum output value for activation.

Parameters:

val - The minimum value.

setMax

public void setMax(double val)

The maximum value for output activation.

Parameters:

val - The maximum value.

getMin

public double getMin()

The minimum value for activation.

Returns:

The minimum value for activation.

getMax

public double getMax()

The maximum value for activation.

Returns:

The max value for activation.

activation

public double activation(double val)

Calculates the activation value as per the sigmoid function.

Specified by:

activation in interface ActivationFunction

Parameters:

val - The raw value.

Returns:

The activation value.

derivative

public double derivative(double val)

Calculate the derivative of the activation value.

Specified by:

derivative in interface ActivationFunction

Parameters:

val - The raw value.

Returns:

The derivative of the raw value.

main

public static void main(java.lang.String[] args)