Same as drwnIdentityJointFeatureMap but adds a bias term for each class i.e., $\phi(x, y) = \left(\delta\!\left\{y = 0\right\} (x^T, 1), \ldots, \delta\!\left\{y = K - 2\right\} (x^T, 1)\right) \in \mathbb{R}^{(K - 1)(n + 1)}$ . More...

Inheritance diagram for drwnBiasJointFeatureMap:

Public Member Functions
	drwnBiasJointFeatureMap (int nFeatures, int nClasses)

const char *	type () const
	returns object type as a string (e.g., Foo::type() { return "Foo"; })

drwnBiasJointFeatureMap *	clone () const
	returns a copy of the class usually implemented as virtual Foo* clone() { return new Foo(*this); }

int	numParameters () const
	returns the number of features in the (joint) output space

void	initialize (int nFeatures, int nClasses)
	initialize number of classes and number of features

vector< double >	operator() (const vector< double > &x, int y) const
	feature vector for given y

vector< double >	operator() (const vector< double > &x) const
	feature vector summed over y

double	dot (const vector< double > &theta, const vector< double > &x, int y) const
	dot product

double	dot (const vector< double > &theta, const vector< double > &x) const
	dot product

void	mac (vector< double > &theta, const vector< double > &x, double alpha, int y) const
	multiply-accumulate

void	mac (vector< double > &theta, const vector< double > &x, const vector< double > &alpha) const
	multiply-accumulate

Public Member Functions inherited from drwnJointFeatureMap
	drwnJointFeatureMap ()
	default constructor

	drwnJointFeatureMap (int nFeatures, int nClasses)
	construct with known number of input features and classes

virtual	~drwnJointFeatureMap ()
	destructor

int	numFeatures () const
	returns the number of features in the input space

int	numClasses () const
	returns the number of classes

double	dot (const vector< double > &theta, const vector< double > &x) const
	returns the dot product $\sum_y \theta^T \phi(x, y)$ (default behaviour is very inefficient)

Protected Attributes
int	_nFeaturesDiv4

int	_nFeaturesMod4

Protected Attributes inherited from drwnJointFeatureMap
int	_nFeatures
	number of input features

int	_nClasses
	number of class labels

Detailed Description

Same as drwnIdentityJointFeatureMap but adds a bias term for each class i.e., $\phi(x, y) = \left(\delta\!\left\{y = 0\right\} (x^T, 1), \ldots, \delta\!\left\{y = K - 2\right\} (x^T, 1)\right) \in \mathbb{R}^{(K - 1)(n + 1)}$ .

The documentation for this class was generated from the following file:

drwnFeatureMaps.h

Public Member Functions

Protected Attributes

Detailed Description