Detailed Description

Class LinearMachine is a generic interface for all kinds of linear machines like classifiers.

A linear classifier computes

$f({\bf x})= {\bf w} \cdot {\bf x} + b$

where ${\bf w}$ are the weights assigned to each feature in training and $b$ the bias.

To implement a linear classifier all that is required is to define the train() function that delivers ${\bf w}$ above.

Note that this framework works with linear classifiers of arbitraty feature type, e.g. dense and sparse and even string based features. This is implemented by using CDotFeatures that may provide a mapping function $\Phi({\bf x})\mapsto {\cal R^D}$ encapsulating all the required operations (like the dot product). The decision function is thus

$f({\bf x})= {\bf w} \cdot \Phi({\bf x}) + b.$

The following linear classifiers are implemented

Linear Descriminant Analysis (CLDA)
Linear Programming Machines (CLPM, CLPBoost)
Perceptron (CPerceptron)
Linear SVMs (CSVMSGD, CLibLinear, CSVMOcas, CSVMLin, CSubgradientSVM)

See also:: CDotFeatures

Definition at line 61 of file LinearMachine.h.

Inheritance diagram for CLinearMachine:

[legend]

List of all members.

Public Member Functions
	CLinearMachine ()
virtual	~CLinearMachine ()
void	get_w (float64_t *&dst_w, int32_t &dst_dims)
SGVector< float64_t >	get_w ()
void	set_w (float64_t *src_w, int32_t src_w_dim)
void	set_bias (float64_t b)
float64_t	get_bias ()
virtual bool	load (FILE *srcfile)
virtual bool	save (FILE *dstfile)
virtual void	set_features (CDotFeatures *feat)
virtual CLabels *	apply ()
virtual CLabels *	apply (CFeatures *data)
virtual float64_t	apply (int32_t vec_idx)
	get output for example "vec_idx"
virtual CDotFeatures *	get_features ()
virtual const char *	get_name () const
Protected Member Functions
virtual void	store_model_features ()
Protected Attributes
int32_t	w_dim
float64_t *	w
float64_t	bias
CDotFeatures *	features

Constructor & Destructor Documentation

CLinearMachine ( )

default constructor

Definition at line 16 of file LinearMachine.cpp.

~CLinearMachine ( ) [virtual]

Definition at line 26 of file LinearMachine.cpp.

Member Function Documentation

CLabels * apply ( ) [virtual]

apply linear machine to all examples

Returns:: resulting labels

Implements CMachine.

Definition at line 46 of file LinearMachine.cpp.

CLabels * apply ( CFeatures * data ) [virtual]

apply linear machine to data

Parameters:

data

(test)data to be classified

Returns:: classified labels

Implements CMachine.

Definition at line 61 of file LinearMachine.cpp.

virtual float64_t apply ( int32_t vec_idx ) [virtual]

get output for example "vec_idx"

Reimplemented from CMachine.

Definition at line 159 of file LinearMachine.h.

float64_t get_bias ( )

get bias

Returns:: bias

Definition at line 115 of file LinearMachine.h.

virtual CDotFeatures* get_features ( ) [virtual]

get features

Returns:: features

Definition at line 168 of file LinearMachine.h.

virtual const char* get_name ( void ) const [virtual]

Returns the name of the SGSerializable instance. It MUST BE the CLASS NAME without the prefixed `C'.

Returns:: name of the SGSerializable

Implements CSGObject.

Reimplemented in CAveragedPerceptron, CLDA, CPerceptron, CDomainAdaptationSVMLinear, CLibLinear, CSGDQN, CSubGradientSVM, CSVMLin, CSVMOcas, and CSVMSGD.

Definition at line 175 of file LinearMachine.h.

void get_w	(	float64_t *&	dst_w,
		int32_t &	dst_dims
	)

get w

Parameters:

	dst_w	store w in this argument
	dst_dims	dimension of w

Definition at line 73 of file LinearMachine.h.

SGVector<float64_t> get_w ( )

get w

Returns:: weight vector

Definition at line 84 of file LinearMachine.h.

bool load ( FILE * srcfile ) [virtual]

load from file

Parameters:

srcfile

file to load from

Returns:: if loading was successful

Reimplemented from CMachine.

Definition at line 32 of file LinearMachine.cpp.

bool save ( FILE * dstfile ) [virtual]

save to file

Parameters:

dstfile

file to save to

Returns:: if saving was successful

Reimplemented from CMachine.

Definition at line 39 of file LinearMachine.cpp.

void set_bias ( float64_t b )

set bias

Parameters:

b

new bias

Definition at line 106 of file LinearMachine.h.

virtual void set_features ( CDotFeatures * feat ) [virtual]

set features

Parameters:

feat

features to set

Reimplemented in CLDA.

Definition at line 138 of file LinearMachine.h.

void set_w	(	float64_t *	src_w,
		int32_t	src_w_dim
	)

set w

Parameters:

	src_w	new w
	src_w_dim	dimension of new w

Definition at line 94 of file LinearMachine.h.

virtual void store_model_features ( ) [protected, virtual]

Stores feature data of underlying model. Does nothing because Linear machines store the normal vector of the separating hyperplane and therefore the model anyway

Reimplemented from CMachine.

Definition at line 182 of file LinearMachine.h.

Member Data Documentation

float64_t bias [protected]

bias

Definition at line 190 of file LinearMachine.h.

CDotFeatures* features [protected]

features

Definition at line 192 of file LinearMachine.h.

float64_t* w [protected]

w

Definition at line 188 of file LinearMachine.h.

int32_t w_dim [protected]

dimension of w

Definition at line 186 of file LinearMachine.h.

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

CLinearMachine Class Reference

Detailed Description

Public Member Functions

Protected Member Functions

Protected Attributes

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation