plusml::MulticlassSVM Class Reference

Class implementing multiclassification SVM. More...

#include <multiclass_svm.h>

Public Member Functions
	MulticlassSVM (uint64_t features, uint64_t classes, ClassificationMode mode, bool bias_enabled=true)
	Constructor for MulticlassSVM.
uint64_t	Features () const
	Get number of features in each sample for the model.
void	FitSGD (const Eigen::MatrixXf &samples, const Eigen::MatrixXi &targets, float learning_rate, float l2_alpha, uint64_t batch_size, uint64_t epochs)
	Fit model using stochastic gradient descent with given hyperparameters.
Eigen::MatrixXi	Predict (Eigen::MatrixXf samples)
	Get predictions for a given matrix of samples.

Detailed Description

Class implementing multiclassification SVM.

Pass plusml::kOneVsOne or plusml::kOneVsAll to the constructor to select classification mode.

Constructor & Destructor Documentation

MulticlassSVM()

plusml::MulticlassSVM::MulticlassSVM	(	uint64_t	features,
		uint64_t	classes,
		ClassificationMode	mode,
		bool	bias_enabled = true )

Constructor for MulticlassSVM.

Parameters

features	Number of features in each sample
classes	Number of classes in classification problem
mode	Classification mode (plusml::kOneVsOne or plusml::kOneVsAll)
bias_enabled	Specifies whether to use bias or not

Member Function Documentation

Features()

uint64_t plusml::MulticlassSVM::Features

(

)

const

Get number of features in each sample for the model.

Returns

Number of features in each sample for the model

FitSGD()

void plusml::MulticlassSVM::FitSGD	(	const Eigen::MatrixXf &	samples,
		const Eigen::MatrixXi &	targets,
		float	learning_rate,
		float	l2_alpha,
		uint64_t	batch_size,
		uint64_t	epochs )

Fit model using stochastic gradient descent with given hyperparameters.

Parameters

samples	Matrix of samples (MxN where M is number of samples and N is number of features in each sample)
targets	Matrix of targets (Mx1 where M is number of samples)
learning_rate	Learning rate for SGD
l2_alpha	Coefficient for L2 regularization
batch_size	Batch size for SGD
epochs	Number of epochs for SGD

Predict()

Eigen::MatrixXi plusml::MulticlassSVM::Predict

(

Eigen::MatrixXf

samples

)

Get predictions for a given matrix of samples.

Parameters

samples

Matrix of samples (MxN where M is number of samples and N is number of features in each sample)

Returns

Matrix of predicted classes (Mx1 where M is number of samples)

---

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

• include/PlusML/multiclass_svm.h
• src/multiclass_svm.cpp