svm_classifier::SVMClassifier Class Reference

Support Vector Machine Classifier with scikit-learn compatible API. More...

#include <svm_classifier.hpp>

Public Member Functions
	SVMClassifier ()
	Default constructor with default parameters.
	SVMClassifier (const nlohmann::json &config)
	Constructor with JSON parameters.
	SVMClassifier (KernelType kernel, double C=1.0, MulticlassStrategy multiclass_strategy=MulticlassStrategy::ONE_VS_REST)
	Constructor with explicit parameters.
	~SVMClassifier ()
	Destructor.
	SVMClassifier (const SVMClassifier &)=delete
	Copy constructor (deleted - models are not copyable)
SVMClassifier &	operator= (const SVMClassifier &)=delete
	Copy assignment (deleted - models are not copyable)
	SVMClassifier (SVMClassifier &&) noexcept
	Move constructor.
SVMClassifier &	operator= (SVMClassifier &&) noexcept
	Move assignment.
TrainingMetrics	fit (const torch::Tensor &X, const torch::Tensor &y)
	Train the SVM classifier.
torch::Tensor	predict (const torch::Tensor &X)
	Predict class labels for samples.
torch::Tensor	predict_proba (const torch::Tensor &X)
	Predict class probabilities for samples.
torch::Tensor	decision_function (const torch::Tensor &X)
	Get decision function values.
double	score (const torch::Tensor &X, const torch::Tensor &y_true)
	Calculate accuracy score on test data.
EvaluationMetrics	evaluate (const torch::Tensor &X, const torch::Tensor &y_true)
	Calculate detailed evaluation metrics.
void	set_parameters (const nlohmann::json &config)
	Set parameters from JSON configuration.
nlohmann::json	get_parameters () const
	Get current parameters as JSON.
bool	is_fitted () const
	Check if the model is fitted/trained.
int	get_n_classes () const
	Get the number of classes.
std::vector< int >	get_classes () const
	Get unique class labels.
int	get_n_features () const
	Get the number of features.
TrainingMetrics	get_training_metrics () const
	Get training metrics from last fit.
bool	supports_probability () const
	Check if the current model supports probability prediction.
void	save_model (const std::string &filename) const
	Save model to file.
void	load_model (const std::string &filename)
	Load model from file.
KernelType	get_kernel_type () const
	Get kernel type.
MulticlassStrategy	get_multiclass_strategy () const
	Get multiclass strategy.
SVMLibrary	get_svm_library () const
	Get SVM library being used.
std::vector< double >	cross_validate (const torch::Tensor &X, const torch::Tensor &y, int cv=5)
	Perform cross-validation.
nlohmann::json	grid_search (const torch::Tensor &X, const torch::Tensor &y, const nlohmann::json &param_grid, int cv=5)
	Find optimal hyperparameters using grid search.
torch::Tensor	get_feature_importance () const
	Get feature importance (for linear kernels only)
void	reset ()
	Reset the classifier (clear trained model)

Detailed Description

Support Vector Machine Classifier with scikit-learn compatible API.

This class provides a unified interface for SVM classification using both liblinear (for linear kernels) and libsvm (for non-linear kernels). It supports multiclass classification through One-vs-Rest and One-vs-One strategies.

Definition at line 21 of file svm_classifier.hpp.

Constructor & Destructor Documentation

SVMClassifier() [1/2]

svm_classifier::SVMClassifier::SVMClassifier ( const nlohmann::json &  config )

explicit

Constructor with JSON parameters.

Parameters

config

JSON configuration object

SVMClassifier() [2/2]

svm_classifier::SVMClassifier::SVMClassifier	(	KernelType	kernel,
		double	C = 1.0,
		MulticlassStrategy	multiclass_strategy = MulticlassStrategy::ONE_VS_REST
	)

Constructor with explicit parameters.

Parameters

kernel	Kernel type
C	Regularization parameter
multiclass_strategy	Multiclass strategy

Member Function Documentation

cross_validate()

std::vector< double > svm_classifier::SVMClassifier::cross_validate	(	const torch::Tensor &	X,
		const torch::Tensor &	y,
		int	cv = 5
	)

Perform cross-validation.

Parameters

X	Feature tensor
y	Target tensor
cv	Number of folds (default: 5)

Returns

Cross-validation scores for each fold

decision_function()

torch::Tensor svm_classifier::SVMClassifier::decision_function

(

const torch::Tensor &

X

)

Get decision function values.

Parameters

X	Feature tensor of shape (n_samples, n_features)

Returns

Tensor with decision function values

Exceptions

std::runtime_error

if model is not fitted

evaluate()

EvaluationMetrics svm_classifier::SVMClassifier::evaluate	(	const torch::Tensor &	X,
		const torch::Tensor &	y_true
	)

Calculate detailed evaluation metrics.

Parameters

X	Feature tensor of shape (n_samples, n_features)
y_true	True labels tensor of shape (n_samples,)

Returns

Evaluation metrics including precision, recall, F1-score

fit()

TrainingMetrics svm_classifier::SVMClassifier::fit	(	const torch::Tensor &	X,
		const torch::Tensor &	y
	)

Train the SVM classifier.

Parameters

X	Feature tensor of shape (n_samples, n_features)
y	Target tensor of shape (n_samples,) with class labels

Returns

Training metrics

Exceptions

std::invalid_argument	if input data is invalid
std::runtime_error	if training fails

get_classes()

std::vector< int > svm_classifier::SVMClassifier::get_classes

(

)

const

Get unique class labels.

Returns

Vector of unique class labels

get_feature_importance()

torch::Tensor svm_classifier::SVMClassifier::get_feature_importance

(

)

const

Get feature importance (for linear kernels only)

Returns

Tensor with feature weights/importance

Exceptions

std::runtime_error

if not supported for current kernel

get_kernel_type()

KernelType svm_classifier::SVMClassifier::get_kernel_type ( ) const

inline

Get kernel type.

Returns

Current kernel type

Definition at line 187 of file svm_classifier.hpp.

get_multiclass_strategy()

MulticlassStrategy svm_classifier::SVMClassifier::get_multiclass_strategy ( ) const

inline

Get multiclass strategy.

Returns

Current multiclass strategy

Definition at line 193 of file svm_classifier.hpp.

get_n_classes()

int svm_classifier::SVMClassifier::get_n_classes

(

)

const

Get the number of classes.

Returns

Number of classes (0 if not fitted)

get_n_features()

int svm_classifier::SVMClassifier::get_n_features ( ) const

inline

Get the number of features.

Returns

Number of features (0 if not fitted)

Definition at line 155 of file svm_classifier.hpp.

get_parameters()

nlohmann::json svm_classifier::SVMClassifier::get_parameters

(

)

const

Get current parameters as JSON.

Returns

JSON object with current parameters

get_svm_library()

SVMLibrary svm_classifier::SVMClassifier::get_svm_library ( ) const

inline

Get SVM library being used.

Returns

SVM library type

Definition at line 199 of file svm_classifier.hpp.

get_training_metrics()

TrainingMetrics svm_classifier::SVMClassifier::get_training_metrics ( ) const

inline

Get training metrics from last fit.

Returns

Training metrics

Definition at line 161 of file svm_classifier.hpp.

grid_search()

nlohmann::json svm_classifier::SVMClassifier::grid_search	(	const torch::Tensor &	X,
		const torch::Tensor &	y,
		const nlohmann::json &	param_grid,
		int	cv = 5
	)

Find optimal hyperparameters using grid search.

Parameters

X	Feature tensor
y	Target tensor
param_grid	JSON object with parameter grid
cv	Number of cross-validation folds

Returns

JSON object with best parameters and score

is_fitted()

bool svm_classifier::SVMClassifier::is_fitted ( ) const

inline

Check if the model is fitted/trained.

Returns

True if model is fitted

Definition at line 137 of file svm_classifier.hpp.

load_model()

void svm_classifier::SVMClassifier::load_model

(

const std::string &

filename

)

Load model from file.

Parameters

filename

Path to load the model from

Exceptions

std::runtime_error

if loading fails

predict()

torch::Tensor svm_classifier::SVMClassifier::predict

(

const torch::Tensor &

X

)

Predict class labels for samples.

Parameters

X	Feature tensor of shape (n_samples, n_features)

Returns

Tensor of predicted class labels

Exceptions

std::runtime_error

if model is not fitted

predict_proba()

torch::Tensor svm_classifier::SVMClassifier::predict_proba

(

const torch::Tensor &

X

)

Predict class probabilities for samples.

Parameters

X	Feature tensor of shape (n_samples, n_features)

Returns

Tensor of shape (n_samples, n_classes) with class probabilities

Exceptions

std::runtime_error

if model is not fitted or doesn't support probabilities

save_model()

void svm_classifier::SVMClassifier::save_model

(

const std::string &

filename

)

const

Save model to file.

Parameters

filename

Path to save the model

Exceptions

std::runtime_error

if saving fails

score()

double svm_classifier::SVMClassifier::score	(	const torch::Tensor &	X,
		const torch::Tensor &	y_true
	)

Calculate accuracy score on test data.

Parameters

X	Feature tensor of shape (n_samples, n_features)
y_true	True labels tensor of shape (n_samples,)

Returns

Accuracy score (fraction of correctly predicted samples)

Exceptions

std::runtime_error

if model is not fitted

set_parameters()

void svm_classifier::SVMClassifier::set_parameters

(

const nlohmann::json &

config

)

Set parameters from JSON configuration.

Parameters

config

JSON configuration object

Exceptions

std::invalid_argument

if parameters are invalid

supports_probability()

bool svm_classifier::SVMClassifier::supports_probability

(

)

const

Check if the current model supports probability prediction.

Returns

True if probabilities are supported

---

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

• include/svm_classifier/svm_classifier.hpp