net.sourceforge.cilib.functions.activation
Class ActivationFunction

java.lang.Object
  net.sourceforge.cilib.functions.Function
      net.sourceforge.cilib.functions.activation.ActivationFunction

All Implemented Interfaces:

Serializable, Differentiable, Cloneable

Direct Known Subclasses:

Sigmoid

public abstract class ActivationFunction
extends Function
implements Differentiable

Activation functions are functions that are typically used within Neurons. This class provides an abstraction for all functions that can be used in this manner.

See Also:

Serialized Form

Constructor Summary

ActivationFunction()

Method Summary

abstract Double	evaluate(Number number) Determine the value of the ActivationFunction at the provided point.
abstract Double	evaluate(Type x) Each function must provide an implementation which returns the function value at the given position.
abstract Double	getGradient(Number number) Determine the gradient of the ActivationFunction at the given point.

Methods inherited from class net.sourceforge.cilib.functions.Function

getBehavioralDomainRegistry, getDimension, getDomain, getDomainRegistry, getMaximum, getMinimum, setBehavioralDomain, setBehaviouralDomainRegistry, setDomain

Methods inherited from class java.lang.Object

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

Methods inherited from interface net.sourceforge.cilib.functions.Differentiable

getGradient

Methods inherited from interface net.sourceforge.cilib.util.Cloneable

getClone

Constructor Detail

ActivationFunction

public ActivationFunction()

Method Detail

evaluate

public abstract Double evaluate(Type x)

Each function must provide an implementation which returns the function value at the given position. The length of the position array should be the same as the function dimension.

Specified by:

evaluate in class Function

Parameters:

x - the position

Returns:

The result of the evaluation.

evaluate

public abstract Double evaluate(Number number)

Determine the value of the ActivationFunction at the provided point. The provided point is simply a Number that is provided as input.

Parameters:

number - The input value.

Returns:

The evaluation of the ActivationFunction, given number as input.

getGradient

public abstract Double getGradient(Number number)

Determine the gradient of the ActivationFunction at the given point.

Parameters:

number - The point at which the gradient is to be determined.

Returns:

The value of the gradient and the provided input.