sgpp::datadriven::BoHyperparameterOptimizer Class Reference

BoHyperparameterOptimizer coordinates data input, fitting and validation modules similarly to SparseGridMiner. More...

#include <BoHyperparameterOptimizer.hpp>

Inheritance diagram for sgpp::datadriven::BoHyperparameterOptimizer:

Public Member Functions
	BoHyperparameterOptimizer (SparseGridMiner *miner, FitterFactory *fitterFactory, DataMiningConfigParser &parser)
	Constructor. More...
	BoHyperparameterOptimizer (const BoHyperparameterOptimizer &rhs)=delete
	Copy constructor deleted - not all members can be copied or cloned . More...
	BoHyperparameterOptimizer (BoHyperparameterOptimizer &&rhs)=default
	Default Move constructor . More...
BoHyperparameterOptimizer &	operator= (BoHyperparameterOptimizer &&rhs)=default
	Default Move assign operator. More...
BoHyperparameterOptimizer &	operator= (const BoHyperparameterOptimizer &rhs)=delete
	Default copy assign operator deleted because not all members can be copied. More...
double	run (bool writeToFile) override
	Run hyperparameter optimization using Bayesian Optimization and random search to warm up. More...
double	transformScore (double original)
	Possible score function transformation to accentuate the optimum. More...
	~BoHyperparameterOptimizer () override=default
	Default destructor. More...
Public Member Functions inherited from sgpp::datadriven::HyperparameterOptimizer
	HyperparameterOptimizer (SparseGridMiner *miner, FitterFactory *fitterFactory, DataMiningConfigParser &parser)
	Constructor. More...
	HyperparameterOptimizer (const HyperparameterOptimizer &rhs)=delete
	Copy constructor deleted - not all members can be copied or cloned . More...
	HyperparameterOptimizer (HyperparameterOptimizer &&rhs)=default
	Default Move constructor . More...
HyperparameterOptimizer &	operator= (HyperparameterOptimizer &&rhs)=default
	Default Move assign operator. More...
HyperparameterOptimizer &	operator= (const HyperparameterOptimizer &rhs)=delete
	Default copy assign operator deleted because not all members can be copied. More...
virtual	~HyperparameterOptimizer ()=default
	Default destructor. More...

Additional Inherited Members
Protected Attributes inherited from sgpp::datadriven::HyperparameterOptimizer
HPOConfig	config
	Configuration for all hpo details. More...
std::unique_ptr< FitterFactory >	fitterFactory
	FitterFactory to provide fitters for running different hyperparameter configurations. More...
std::unique_ptr< SparseGridMiner >	miner
	Miner providing all testing facilities. More...

Detailed Description

BoHyperparameterOptimizer coordinates data input, fitting and validation modules similarly to SparseGridMiner.

It offers access to different hyperparameter optimization procedures.

Constructor & Destructor Documentation

BoHyperparameterOptimizer() [1/3]

sgpp::datadriven::BoHyperparameterOptimizer::BoHyperparameterOptimizer	(	SparseGridMiner *	miner,
		FitterFactory *	fitterFactory,
		DataMiningConfigParser &	parser
	)

Constructor.

Parameters

miner	configured instance of SGMiner object, that will provide the learning process. The HyperparameterOptimizer instance will take ownership of the passed object.
fitterFactory	configured instance of factory object that provides fitters with manipulated hyperparameters. The HyperparameterOptimizer instance will take ownership of the passed object.
parser	reference to parser object to read configuration info

BoHyperparameterOptimizer() [2/3]

sgpp::datadriven::BoHyperparameterOptimizer::BoHyperparameterOptimizer ( const BoHyperparameterOptimizer &  rhs )

delete

Copy constructor deleted - not all members can be copied or cloned .

Parameters

rhs	the object to copy from

BoHyperparameterOptimizer() [3/3]

sgpp::datadriven::BoHyperparameterOptimizer::BoHyperparameterOptimizer ( BoHyperparameterOptimizer &&  rhs )

default

Default Move constructor .

Parameters

rhs	the object to move from

~BoHyperparameterOptimizer()

sgpp::datadriven::BoHyperparameterOptimizer::~BoHyperparameterOptimizer ( )

overridedefault

Default destructor.

Member Function Documentation

operator=() [1/2]

BoHyperparameterOptimizer& sgpp::datadriven::BoHyperparameterOptimizer::operator= ( BoHyperparameterOptimizer &&  rhs )

default

Default Move assign operator.

Parameters

rhs	the object to move from

operator=() [2/2]

BoHyperparameterOptimizer& sgpp::datadriven::BoHyperparameterOptimizer::operator= ( const BoHyperparameterOptimizer &  rhs )

delete

Default copy assign operator deleted because not all members can be copied.

Parameters

rhs	the object to copy from

run()

double sgpp::datadriven::BoHyperparameterOptimizer::run ( bool  writeToFile )

overridevirtual

Run hyperparameter optimization using Bayesian Optimization and random search to warm up.

Implements sgpp::datadriven::HyperparameterOptimizer.

References sgpp::datadriven::HyperparameterOptimizer::config, sgpp::optimization::Printer::disableStatusPrinting(), sgpp::datadriven::BayesianOptimization::fitScales(), sgpp::datadriven::HyperparameterOptimizer::fitterFactory, sgpp::optimization::Printer::getInstance(), sgpp::datadriven::HPOConfig::getNRandom(), sgpp::datadriven::HPOConfig::getNRuns(), sgpp::datadriven::HPOConfig::getSeed(), python.statsfileInfo::i, sgpp::datadriven::BayesianOptimization::main(), sgpp::datadriven::HyperparameterOptimizer::miner, sgpp::datadriven::BayesianOptimization::setScales(), sgpp::datadriven::BOConfig::setScore(), transformScore(), and sgpp::datadriven::BayesianOptimization::updateGP().

transformScore()

double sgpp::datadriven::BoHyperparameterOptimizer::transformScore

(

double

original

)

Possible score function transformation to accentuate the optimum.

Parameters

original

Returns

transformed value

Referenced by run().

---

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

• datadriven/src/sgpp/datadriven/datamining/modules/hpo/BoHyperparameterOptimizer.hpp
• datadriven/src/sgpp/datadriven/datamining/modules/hpo/BoHyperparameterOptimizer.cpp