This provides an alternative to using GActivationIdentity on the output layer for regression problems. It may add more power because it is non-linear, but like the identity function, its co-domain is the same as its domain. At very positive values, this is shaped like y=1.5*x. At very negative values, this is shaped like y=0.5*x. Around 0, it is shaped like y=x. More...

#include <GActivation.h>

Inheritance diagram for GClasses::GActivationBend:

Public Member Functions
virtual double	center ()
	Returns 0.0. More...

virtual GActivationFunction *	clone ()
	See the comment for GActivationFunction::clone. More...

virtual double	derivative (double x)
	Returns the derivative of the bend function. More...

virtual double	halfRange ()
	Returns 10000. More...

virtual double	identityBias ()
	Returns 0.0. More...

virtual double	identityDiag ()
	Returns 1.0. More...

virtual double	inverse (double y)
	Returns the inverse of the bend function. More...

virtual const char *	name () const
	Returns the name of this activation function. More...

virtual double	squash (double x)
	Returns the bend function of x. More...

Public Member Functions inherited from GClasses::GActivationFunction
	GActivationFunction ()

virtual	~GActivationFunction ()

virtual double	derivativeOfNet (double net, double activation)
	This computes the derivative of the net value. (Sometimes, such as with GActivationLogistic, it is more efficient to compute this from the activation value, so both are provided.) More...

GDomNode *	serialize (GDom *pDoc) const
	Serialize this object. More...

Additional Inherited Members
Static Public Member Functions inherited from GClasses::GActivationFunction
static GActivationFunction *	deserialize (GDomNode *pNode)
	Deserialize this object. More...

Detailed Description

This provides an alternative to using GActivationIdentity on the output layer for regression problems. It may add more power because it is non-linear, but like the identity function, its co-domain is the same as its domain. At very positive values, this is shaped like y=1.5*x. At very negative values, this is shaped like y=0.5*x. Around 0, it is shaped like y=x.