SGD Class Reference

#include <SGD.h>

Public Member Functions
	SGD (double _eta=0.01, size_t iter=10, bool _shuffle=false, const std::function< double(const Matrix< double > &, const Matrix< double > &)> &update_weights_fun=nullptr, const std::function< Matrix< double >(const Matrix< double > &)> &netInputFun=nullptr)
void	fit (const Matrix< double > &X, const Matrix< double > &y, Matrix< double > &weights)
void	partial_fit (const Matrix< double > &X, const Matrix< double > &y, Matrix< double > &weights) const
double	update_weights (const Matrix< double > &xi, const Matrix< double > &target, Matrix< double > &weights) const
std::pair< Matrix< double >, Matrix< double > >	shuffleData (const Matrix< double > &X, const Matrix< double > &y)

Static Public Member Functions
static Matrix< double >	netInput (const Matrix< double > &X, const Matrix< double > &weights)

Public Attributes
Matrix< double >	cost
	matrix holding cost per epoch More...

Private Attributes
size_t	n_iter
	number iterations (epochs) during fit More...
double	eta
	cost factor More...
bool	shuffle = false
	shuffle during training More...
std::function< double(const Matrix< double > &, const Matrix< double > &)>	weight_update = nullptr
	represents weight update function More...
std::function< Matrix< double >(const Matrix< double > &)>	net_input_fun = nullptr
	represents net input function More...

Detailed Description

Implements statistic gradient decent method to fit a model

Constructor & Destructor Documentation

SGD()

SGD::SGD ( double  _eta = 0.01, size_t  iter = 10, bool  _shuffle = false, const std::function< double(const Matrix< double > &, const Matrix< double > &)> &  update_weights_fun = nullptr, const std::function< Matrix< double >(const Matrix< double > &)> &  netInputFun = nullptr  )

inlineexplicit

default constructor

Parameters

_eta
iter
_shuffle
update_weights_fun
netInputFun

Member Function Documentation

fit()

void SGD::fit ( const Matrix< double > &  X, const Matrix< double > &  y, Matrix< double > &  weights  )

inline

fits given weights according to used weight update/net input function based on given values X and y

Parameters

X
y
weights

netInput()

static Matrix< double > SGD::netInput ( const Matrix< double > &  X, const Matrix< double > &  weights  )

inlinestatic

computes the net-input for given values

Parameters

X
weights

Returns

partial_fit()

void SGD::partial_fit ( const Matrix< double > &  X, const Matrix< double > &  y, Matrix< double > &  weights  ) const

inline

Performs partial fit of given weights to input values

Parameters

X
y
weights

shuffleData()

std::pair< Matrix< double >, Matrix< double > > SGD::shuffleData ( const Matrix< double > &  X, const Matrix< double > &  y  )

inline

shuffles given data

Parameters

X
y

Returns

update_weights()

double SGD::update_weights ( const Matrix< double > &  xi, const Matrix< double > &  target, Matrix< double > &  weights  ) const

inline

calculates update values (mean-square-error)

Parameters

xi
target
weights

Returns

Member Data Documentation

cost

Matrix<double> SGD::cost

matrix holding cost per epoch

eta

double SGD::eta

private

cost factor

n_iter

size_t SGD::n_iter

private

number iterations (epochs) during fit

net_input_fun

std::function<Matrix<double>(const Matrix<double>&)> SGD::net_input_fun = nullptr

private

represents net input function

shuffle

bool SGD::shuffle = false

private

shuffle during training

weight_update

std::function<double(const Matrix<double>&, const Matrix<double>&)> SGD::weight_update = nullptr

private

represents weight update function

---

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

• include/math/ds/SGD.h