Detailed Description

The CommWordString kernel may be used to compute the spectrum kernel from strings that have been mapped into unsigned 16bit integers.

These 16bit integers correspond to k-mers. To applicable in this kernel they need to be sorted (e.g. via the SortWordString pre-processor).

It basically uses the algorithm in the unix "comm" command (hence the name) to compute:

$k({\bf x},({\bf x'})= \Phi_k({\bf x})\cdot \Phi_k({\bf x'})$

where $\Phi_k$ maps a sequence ${\bf x}$ that consists of letters in $\Sigma$ to a feature vector of size $|\Sigma|^k$ . In this feature vector each entry denotes how often the k-mer appears in that ${\bf x}$ .

Note that this representation is especially tuned to small alphabets (like the 2-bit alphabet DNA), for which it enables spectrum kernels of order up to 8.

For this kernel the linadd speedups are quite efficiently implemented using direct maps.

Definition at line 46 of file CommWordStringKernel.h.

Inheritance diagram for CCommWordStringKernel:

[legend]

List of all members.

Public Member Functions
	CCommWordStringKernel ()
	CCommWordStringKernel (int32_t size, bool use_sign)
	CCommWordStringKernel (CStringFeatures< uint16_t > l, CStringFeatures< uint16_t > r, bool use_sign=false, int32_t size=10)
virtual	~CCommWordStringKernel ()
virtual bool	init (CFeatures l, CFeatures r)
virtual void	cleanup ()
virtual EKernelType	get_kernel_type ()
virtual const char *	get_name () const
virtual bool	init_dictionary (int32_t size)
virtual bool	init_optimization (int32_t count, int32_t IDX, float64_t weights)
virtual bool	delete_optimization ()
virtual float64_t	compute_optimized (int32_t idx)
virtual void	add_to_normal (int32_t idx, float64_t weight)
virtual void	clear_normal ()
virtual EFeatureType	get_feature_type ()
void	get_dictionary (int32_t &dsize, float64_t *&dweights)
virtual float64_t *	compute_scoring (int32_t max_degree, int32_t &num_feat, int32_t &num_sym, float64_t target, int32_t num_suppvec, int32_t IDX, float64_t *alphas, bool do_init=true)
char *	compute_consensus (int32_t &num_feat, int32_t num_suppvec, int32_t IDX, float64_t alphas)
void	set_use_dict_diagonal_optimization (bool flag)
bool	get_use_dict_diagonal_optimization ()
Protected Member Functions
virtual float64_t	compute (int32_t idx_a, int32_t idx_b)
virtual float64_t	compute_helper (int32_t idx_a, int32_t idx_b, bool do_sort)
virtual float64_t	compute_diag (int32_t idx_a)
Protected Attributes
int32_t	dictionary_size
float64_t *	dictionary_weights
bool	use_sign
bool	use_dict_diagonal_optimization
int32_t *	dict_diagonal_optimization
Friends
class	CVarianceKernelNormalizer
class	CSqrtDiagKernelNormalizer
class	CAvgDiagKernelNormalizer
class	CRidgeKernelNormalizer
class	CFirstElementKernelNormalizer
class	CTanimotoKernelNormalizer
class	CDiceKernelNormalizer

Constructor & Destructor Documentation

CCommWordStringKernel ( )

default constructor

Definition at line 22 of file CommWordStringKernel.cpp.

CCommWordStringKernel	(	int32_t	size,
		bool	use_sign
	)

constructor

Parameters:

	size	cache size
	use_sign	if sign shall be used

Definition at line 28 of file CommWordStringKernel.cpp.

CCommWordStringKernel	(	CStringFeatures< uint16_t > *	l,
		CStringFeatures< uint16_t > *	r,
		bool	use_sign = `false`,
		int32_t	size = `10`
	)

constructor

Parameters:

	l	features of left-hand side
	r	features of right-hand side
	use_sign	if sign shall be used
	size	cache size

Definition at line 35 of file CommWordStringKernel.cpp.

~CCommWordStringKernel ( ) [virtual]

Definition at line 57 of file CommWordStringKernel.cpp.

Member Function Documentation

void add_to_normal	(	int32_t	idx,
		float64_t	weight
	)			`[virtual]`

add to normal

Parameters:

	idx	where to add
	weight	what to add

Reimplemented from CKernel.

Reimplemented in CWeightedCommWordStringKernel.

Definition at line 241 of file CommWordStringKernel.cpp.

void cleanup ( ) [virtual]

clean up kernel

Reimplemented from CKernel.

Reimplemented in CWeightedCommWordStringKernel.

Definition at line 79 of file CommWordStringKernel.cpp.

void clear_normal ( ) [virtual]

clear normal

Reimplemented from CKernel.

Definition at line 286 of file CommWordStringKernel.cpp.

virtual float64_t compute	(	int32_t	idx_a,
		int32_t	idx_b
	)			`[protected, virtual]`

compute kernel function for features a and b idx_{a,b} denote the index of the feature vectors in the corresponding feature object

Parameters:

	idx_a	index a
	idx_b	index b

Returns:: computed kernel function at indices a,b

Implements CKernel.

Definition at line 215 of file CommWordStringKernel.h.

char * compute_consensus	(	int32_t &	num_feat,
		int32_t	num_suppvec,
		int32_t *	IDX,
		float64_t *	alphas
	)

compute consensus

Parameters:

	num_feat	number of features
	num_suppvec	number of support vectors
	IDX	IDX
	alphas	alphas

Returns:: computed consensus

Definition at line 498 of file CommWordStringKernel.cpp.

float64_t compute_diag ( int32_t idx_a ) [protected, virtual]

helper to compute only diagonal normalization for training

Parameters:

idx_a

index a

Returns:: unnormalized diagonal value

Definition at line 85 of file CommWordStringKernel.cpp.

float64_t compute_helper	(	int32_t	idx_a,
		int32_t	idx_b,
		bool	do_sort
	)			`[protected, virtual]`

helper for compute

Parameters:

	idx_a	index a
	idx_b	index b
	do_sort	if sorting shall be performed

Returns:: computed value

Reimplemented in CWeightedCommWordStringKernel.

Definition at line 129 of file CommWordStringKernel.cpp.

float64_t compute_optimized ( int32_t idx ) [virtual]

compute optimized

Parameters:

idx

index to compute

Returns:: optimized value at given index

Reimplemented from CKernel.

Reimplemented in CWeightedCommWordStringKernel.

Definition at line 326 of file CommWordStringKernel.cpp.

float64_t * compute_scoring	(	int32_t	max_degree,
		int32_t &	num_feat,
		int32_t &	num_sym,
		float64_t *	target,
		int32_t	num_suppvec,
		int32_t *	IDX,
		float64_t *	alphas,
		bool	do_init = `true`
	)			`[virtual]`

compute scoring

Parameters:

	max_degree	maximum degree
	num_feat	number of features
	num_sym	number of symbols
	target	target
	num_suppvec	number of support vectors
	IDX	IDX
	alphas	alphas
	do_init	if initialization shall be performed