krotos::LinearModel Class Reference

class that infers linear model it supports multiple fully connected hidden layers and calculates multiple outputs focusing on classification It features an optional sigmoid and a relu activation function More...

#include <LinearModel.h>

Public Member Functions
	LinearModel (const std::vector< Eigen::MatrixXf > &weights, const std::vector< Eigen::VectorXf > &biases, const Eigen::MatrixXf &weightsOutput, const Eigen::VectorXf &biasOutput, const StringArray &categories)
	Constructor for the linear model this will be called to set manually the parametrs of the model.
	LinearModel ()
	Constructor for the linear model, this will automatically load a model that is expected to exist in the binary data.
Eigen::VectorXf	infer (const Eigen::VectorXf &input)
	infers the model to predict output
Eigen::VectorXf	inferSingle (const Eigen::VectorXf &input)
String	getProminentCategory ()
	returns the prominent category name
StringArray	getOutputCategories ()
	returns the output categories
void	loadModelData ()

Private Member Functions
Eigen::VectorXf	sigmoid (const Eigen::VectorXf &lOut)
	sigmoid activation function
Eigen::VectorXf	relu (const Eigen::VectorXf &lOut)
	relu activation function

Private Attributes
std::vector< Eigen::MatrixXf >	m_weights
Eigen::MatrixXf	m_weightsOutput
std::vector< Eigen::VectorXf >	m_bias
Eigen::VectorXf	m_biasOutput
StringArray	m_outputCategories
std::vector< Eigen::VectorXf >	m_layerOutput
Eigen::VectorXf	m_outputs
int	m_numLayers {0}
int	m_layerSize {0}
int	m_outputSize {0}
var	m_weightsVar
Eigen::VectorXf	m_workingBuffer
int	m_prominentCategoryIndex = {0}

Detailed Description

class that infers linear model it supports multiple fully connected hidden layers and calculates multiple outputs focusing on classification It features an optional sigmoid and a relu activation function

Constructor & Destructor Documentation

LinearModel() [1/2]

krotos::LinearModel::LinearModel	(	const std::vector< Eigen::MatrixXf > &	weights,
		const std::vector< Eigen::VectorXf > &	biases,
		const Eigen::MatrixXf &	weightsOutput,
		const Eigen::VectorXf &	biasOutput,
		const StringArray &	categories )

Constructor for the linear model this will be called to set manually the parametrs of the model.

Parameters

w	the weights vector
b	layer bias

LinearModel() [2/2]

krotos::LinearModel::LinearModel

(

)

Constructor for the linear model, this will automatically load a model that is expected to exist in the binary data.

Member Function Documentation

getOutputCategories()

StringArray krotos::LinearModel::getOutputCategories

(

)

returns the output categories

getProminentCategory()

String krotos::LinearModel::getProminentCategory

(

)

returns the prominent category name

infer()

Eigen::VectorXf krotos::LinearModel::infer

(

const Eigen::VectorXf &

input

)

infers the model to predict output

Parameters

input

the input data to predict

Returns

output scores for each category

inferSingle()

Eigen::VectorXf krotos::LinearModel::inferSingle

(

const Eigen::VectorXf &

input

)

loadModelData()

void krotos::LinearModel::loadModelData

(

)

@rief loads the model data from binary data and sets up the relevant variables

relu()

Eigen::VectorXf krotos::LinearModel::relu ( const Eigen::VectorXf & lOut )

private

relu activation function

sigmoid()

Eigen::VectorXf krotos::LinearModel::sigmoid ( const Eigen::VectorXf & lOut )

private

sigmoid activation function

Member Data Documentation

m_bias

std::vector<Eigen::VectorXf> krotos::LinearModel::m_bias

private

m_biasOutput

Eigen::VectorXf krotos::LinearModel::m_biasOutput

private

m_layerOutput

std::vector<Eigen::VectorXf> krotos::LinearModel::m_layerOutput

private

m_layerSize

int krotos::LinearModel::m_layerSize {0}

private

m_numLayers

int krotos::LinearModel::m_numLayers {0}

private

m_outputCategories

StringArray krotos::LinearModel::m_outputCategories

private

m_outputs

Eigen::VectorXf krotos::LinearModel::m_outputs

private

m_outputSize

int krotos::LinearModel::m_outputSize {0}

private

m_prominentCategoryIndex

int krotos::LinearModel::m_prominentCategoryIndex = {0}

private

m_weights

std::vector<Eigen::MatrixXf> krotos::LinearModel::m_weights

private

m_weightsOutput

Eigen::MatrixXf krotos::LinearModel::m_weightsOutput

private

m_weightsVar

var krotos::LinearModel::m_weightsVar

private

m_workingBuffer

Eigen::VectorXf krotos::LinearModel::m_workingBuffer

private

---

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

• krotos_ml/NLP/LinearModel.h
• krotos_ml/NLP/LinearModel.cpp