meta::classify Namespace Reference

Algorithms for feature selection, KNN search, and confusion matrices. More...

Namespaces
	kernel
	Kernel functions for linear classifiers.
	loss
	Loss functions for sgd.

Classes
class	binary_classifier
	A classifier which classifies documents as "positive" or "negative". More...
class	binary_classifier_factory
	Factory that is responsible for creating binary classifiers from configuration files. More...
class	classifier
	A classifier uses a document's feature space to identify which group it belongs to. More...
class	classifier_factory
	Factory that is responsible for creating classifiers from configuration files. More...
class	confusion_matrix
	Allows interpretation of classification errors. More...
class	dual_perceptron
	Implements a perceptron classifier, but using the dual formulation of the problem. More...
class	knn
	Implements the k-Nearest Neighbor lazy learning classification algorithm. More...
class	linear_model
	A storage class for multiclass linear classifier models. More...
class	linear_model_exception
	Exception thrown during interactions with linear_models. More...
class	logistic_regression
	Multinomial logistic regression. More...
class	naive_bayes
	Implements the Naive Bayes classifier, a simplistic probabilistic classifier that uses Bayes' theorem with strong feature independence assumptions. More...
class	nearest_centroid
	Implements the nearest centroid classification algorithm. More...
class	one_vs_all
	Generalizes binary classifiers to operate over multiclass types using the one vs all method. More...
class	one_vs_one
	Ensemble method adaptor for extending binary_classifiers to the multi-class classification case by using a one-vs-one strategy. More...
class	sgd
	Implements stochastic gradient descent for learning binary linear classifiers. More...
class	svm_wrapper
	Wrapper class for liblinear (http://www.csie.ntu.edu.tw/~cjlin/liblinear/) and libsvm (http://www.csie.ntu.edu.tw/~cjlin/libsvm/) implementation of support vector machine classification. More...
class	winnow
	Implements the Winnow classifier, a simplistic linear classifier for linearly-separable data. More...

Functions
template<class Index , class Classifier >
void	batch_train (Index &idx, Classifier &cls, const std::vector< doc_id > &training_set, uint64_t batch_size)
	This trains a classifier in an online fashion, using batches of size batch_size from the training_set. More...
std::unique_ptr< binary_classifier >	make_binary_classifier (const cpptoml::table &config, std::shared_ptr< index::forward_index > idx, class_label positive, class_label negative)
	(Non-template): Convenience method for creating a binary classifier using the factory. More...
template<class Classifier >
void	register_binary_classifier ()
	Registration method for binary classifiers. More...
template<>
std::unique_ptr< classifier >	make_classifier< dual_perceptron > (const cpptoml::table &, std::shared_ptr< index::forward_index >)
	Specialization of the factory function used to create dual_perceptrons.
template<>
std::unique_ptr< classifier >	make_multi_index_classifier< knn > (const cpptoml::table &, std::shared_ptr< index::forward_index >, std::shared_ptr< index::inverted_index >)
	Specialization of the factory method used to create knn classifiers.
template<>
std::unique_ptr< classifier >	make_classifier< logistic_regression > (const cpptoml::table &, std::shared_ptr< index::forward_index >)
	Specialization of the factory method used for creating logistic_regression classifiers.
template<>
std::unique_ptr< classifier >	make_classifier< naive_bayes > (const cpptoml::table &config, std::shared_ptr< index::forward_index > idx)
	Specialization of the factory method used for creating naive bayes classifiers.
template<>
std::unique_ptr< classifier >	make_multi_index_classifier< nearest_centroid > (const cpptoml::table &, std::shared_ptr< index::forward_index >, std::shared_ptr< index::inverted_index >)
	Specialization of the factory method used to create nearest_centroid classifiers.
template<>
std::unique_ptr< classifier >	make_classifier< one_vs_all > (const cpptoml::table &, std::shared_ptr< index::forward_index >)
	Specialization of the factory method used to create one_vs_all classifiers.
template<>
std::unique_ptr< classifier >	make_classifier< one_vs_one > (const cpptoml::table &, std::shared_ptr< index::forward_index >)
	Specialization of the factory method used to create one_vs_all classifiers.
template<>
std::unique_ptr< binary_classifier >	make_binary_classifier< sgd > (const cpptoml::table &config, std::shared_ptr< index::forward_index > idx, class_label positive, class_label negative)
	Specialization of the factory method used to create sgd classifiers.
template<>
std::unique_ptr< classifier >	make_classifier< svm_wrapper > (const cpptoml::table &, std::shared_ptr< index::forward_index >)
	Specialization of the factory method used for creating svm_wrapper classifiers.
template<>
std::unique_ptr< classifier >	make_classifier< winnow > (const cpptoml::table &config, std::shared_ptr< index::forward_index > idx)
	Specialization of the factory method used for creating winnow classifiers.
std::unique_ptr< classifier >	make_classifier (const cpptoml::table &config, std::shared_ptr< index::forward_index > idx, std::shared_ptr< index::inverted_index > inv_idx=nullptr)
	Convenience method for creating a classifier using the factory. More...
template<class Classifier >
std::unique_ptr< classifier >	make_classifier (const cpptoml::table &, std::shared_ptr< index::forward_index > idx)
	Factory method for creating a classifier. More...
template<class Classifier >
std::unique_ptr< classifier >	make_multi_index_classifier (const cpptoml::table &, std::shared_ptr< index::forward_index > idx, std::shared_ptr< index::inverted_index > inv_idx)
	Factory method for creating a classifier that takes both index types. More...
template<class Classifier >
void	register_classifier ()
	Registration method for classifiers. More...
template<class Classifier >
void	register_multi_index_classifier ()
	Registration method for multi-index classifiers. More...
term_probs_	reserve (lbls.size())
	for (const auto &lbl:lbls) term_probs_.emplace_back(lbl

Variables
auto	lbls = idx_->class_labels()
	term_prior

Detailed Description

Algorithms for feature selection, KNN search, and confusion matrices.

Function Documentation

template<class Index , class Classifier >

void meta::classify::batch_train	(	Index &	idx,
		Classifier &	cls,
		const std::vector< doc_id > &	training_set,
		uint64_t	batch_size
	)

This trains a classifier in an online fashion, using batches of size batch_size from the training_set.

Parameters

idx	The index the classifier is using (so the cache may be dropped between batches)
cls	The classifier to train. This must be a classifier supporting online learning (e.g., sgd or an ensemble of sgd)
training_set	The list of document ids that comprise the training data
batch_size	The size of the batches to use for the minibatch training

std::unique_ptr< binary_classifier > meta::classify::make_binary_classifier	(	const cpptoml::table &	config,
		std::shared_ptr< index::forward_index >	idx,
		class_label	positive,
		class_label	negative
	)

(Non-template): Convenience method for creating a binary classifier using the factory.

(Template): Factory method for creating a binary classifier; this should be specialized if your given binary classifier requires special construction behavior (e.g., reading parameters).

Parameters

config	The table that specifies the binary classifier's configuration
idx	The forward_index the binary classifier is being constructed over
positive	The class_label for positive documents
negative	The class_label for negative documents

Returns

a unique_ptr to a binary_classifier constructed from the given configuration

Parameters

config	The table that specifies the binary classifier's configuration
idx	The forward_index the binary classifier is being constructed over
positive	The class_label for positive documents
negative	The class_label for negative documents

Returns

a unique_ptr to a binary_classifier (of derived type Classifier) that has been constructed from the given configuration

template<class Classifier >

void meta::classify::register_binary_classifier

(

)

Registration method for binary classifiers.

Clients should use this method to register any new binary classifiers they write.

std::unique_ptr< classifier > meta::classify::make_classifier	(	const cpptoml::table &	config,
		std::shared_ptr< index::forward_index >	idx,
		std::shared_ptr< index::inverted_index >	inv_idx = nullptr
	)

Convenience method for creating a classifier using the factory.

Parameters

config	The configuration group that specifies the configuration for the classifier to be created
idx	The forward_index to be passed to the classifier being created
inv_idx	The inverted_index to be passed to the classifier being created (if needed)

Returns

a unique_ptr to the classifier created from the given configuration

template<class Classifier >

std::unique_ptr<classifier> meta::classify::make_classifier	(	const cpptoml::table &	,
		std::shared_ptr< index::forward_index >	idx
	)

Factory method for creating a classifier.

This should be specialized if your given classifier requires special construction behavior (e.g., reading parameters).

Parameters

config	The configuration group that specifies the configuration for the classifier to be created
idx	The forward_index to be passed to the classifier being created

Returns

a unique_ptr to the classifier (of derived type Classifier) created from the given configuration

template<class Classifier >

std::unique_ptr<classifier> meta::classify::make_multi_index_classifier	(	const cpptoml::table &	,
		std::shared_ptr< index::forward_index >	idx,
		std::shared_ptr< index::inverted_index >	inv_idx
	)

Factory method for creating a classifier that takes both index types.

This should be specialized if your given classifier requires special construction behavior.

Parameters

config	The configuration group that specifies the configuration for the classifier to be created
idx	The forward_index to be passed to the classifier being created
inv_idx	The inverted_index to be passed to the classifier being created

Returns

a unique_ptr to the classifier (of derived type Classifier) created from the given configuration

template<class Classifier >

void meta::classify::register_classifier

(

)

Registration method for classifiers.

Clients should use this method to register any new classifiers they write that operate on just a forward_index (this should be most).

template<class Classifier >

void meta::classify::register_multi_index_classifier

(

)

Registration method for multi-index classifiers.

Clients should use this method to register any new classifiers they write that operate on both a forward_index and an inverted_index (this is rare).