SG++-Doxygen-Documentation
sgpp::datadriven::BayesianOptimization Class Reference

Class to host all methods to perform Bayesian Optimization. More...

#include <BayesianOptimization.hpp>

Public Member Functions

double acquisitionOuter (const base::DataVector &inp)
 Wrapper function for use in optimizer. More...
 
 BayesianOptimization (const std::vector< BOConfig > &initialConfigs)
 Constructor setting up Gaussian Process. More...
 
void decomposeCholesky (base::DataMatrix &km, base::DataMatrix &gnew)
 Perform a Cholesky Decomposition. More...
 
base::DataVector fitScales ()
 fitting the Gaussian Process to the data in the form of scaling the hyperparameters in relation to each other by likelihood maximization More...
 
double kernel (double distance)
 kernel function More...
 
double likelihood (const base::DataVector &inp)
 Gaussian Process likelihood to perform likelihood maximization over. More...
 
BOConfig main (BOConfig &prototype)
 main routine to find new sample point More...
 
double mean (base::DataVector &knew)
 
void setScales (base::DataVector nscales, double factor)
 
void solveCholeskySystem (base::DataMatrix &gmatrix, base::DataVector &x)
 Solve a system of linear equations using previously decomposed matrix. More...
 
void updateGP (BOConfig &newConfig, bool normalize)
 Gaussian Process update step. More...
 
double var (base::DataVector &knew, double kself)
 

Static Public Member Functions

static double acquisitionEI (double dMean, double dVar, double bestsofar)
 Implementation of mathematical formulation of the expected improvement acquisition function. More...
 

Protected Attributes

std::vector< BOConfigallConfigs
 existing sample points in the Gaussian Process More...
 
double bestsofar
 best score encountered so far (used for expected improvement calculation) More...
 
bool decomFailed = false
 debugging variable for numerical instabilities More...
 
base::DataMatrix gleft
 Cholesky Decomposition of the Gram matrix. More...
 
base::DataMatrix kernelmatrix
 Gram matrix containing all kernel values between all existing samples. More...
 
double maxofmax
 debugging variable for numerical instabilities More...
 
base::DataVector rawScores
 score values belonging to all existing sample points More...
 
base::DataVector scales
 scales of the hyperparameter space used to compute the kernel (set by fitScales()) More...
 
bool screwedvar
 debugging variable for numerical instabilities More...
 
base::DataVector transformedOutput
 solution of Kx = s where K is the Gram matrix and s the scores of the samples More...
 

Detailed Description

Class to host all methods to perform Bayesian Optimization.

Constructor & Destructor Documentation

sgpp::datadriven::BayesianOptimization::BayesianOptimization ( const std::vector< BOConfig > &  initialConfigs)
explicit

Constructor setting up Gaussian Process.

Parameters
initialConfigsnon-empty vector of initial points to build the Gaussian Process

References allConfigs, bestsofar, python.statsfileInfo::i, sgpp::base::DataVector::max(), sgpp::base::DataVector::min(), sgpp::base::DataVector::normalize(), rawScores, scales, setScales(), sgpp::base::DataVector::sub(), and sgpp::base::DataVector::sum().

Member Function Documentation

double sgpp::datadriven::BayesianOptimization::acquisitionEI ( double  dMean,
double  dVar,
double  bestsofar 
)
static

Implementation of mathematical formulation of the expected improvement acquisition function.

Referenced by acquisitionOuter().

double sgpp::datadriven::BayesianOptimization::acquisitionOuter ( const base::DataVector inp)

Wrapper function for use in optimizer.

Parameters
inppoint in continuous optimization space
Returns
score to optimize on

References acquisitionEI(), allConfigs, bestsofar, python.statsfileInfo::i, kernel(), m, mean(), scales, and var().

Referenced by main().

void sgpp::datadriven::BayesianOptimization::decomposeCholesky ( base::DataMatrix km,
base::DataMatrix gnew 
)

Perform a Cholesky Decomposition.

Parameters
kminput matrix
gnewoutput (triangular) matrix

References decomFailed, sgpp::base::DataMatrix::get(), sgpp::base::DataMatrix::getNrows(), python.statsfileInfo::i, python.utils.statsfile2gnuplot::j, and sgpp::base::DataMatrix::set().

Referenced by likelihood(), setScales(), and updateGP().

base::DataVector sgpp::datadriven::BayesianOptimization::fitScales ( )

fitting the Gaussian Process to the data in the form of scaling the hyperparameters in relation to each other by likelihood maximization

References likelihood(), and scales.

Referenced by sgpp::datadriven::BoHyperparameterOptimizer::run().

double sgpp::datadriven::BayesianOptimization::kernel ( double  distance)

kernel function

Parameters
distanceas computed according to some spacial representation
Returns
kernel value representing variance

Referenced by acquisitionOuter(), likelihood(), setScales(), and updateGP().

double sgpp::datadriven::BayesianOptimization::likelihood ( const base::DataVector inp)

Gaussian Process likelihood to perform likelihood maximization over.

Parameters
inpvector containing scales of the hyperparameter space
Returns
value representative of the likelihood

References allConfigs, decomposeCholesky(), sgpp::base::DataVector::dotProduct(), sgpp::base::DataMatrix::get(), python.statsfileInfo::i, kernel(), sgpp::combigrid::pow(), rawScores, solveCholeskySystem(), and analyse_erg::tmp.

Referenced by fitScales().

void sgpp::datadriven::BayesianOptimization::solveCholeskySystem ( base::DataMatrix gmatrix,
base::DataVector x 
)

Solve a system of linear equations using previously decomposed matrix.

Parameters
gmatrixdecomposed matrix
xtarget vector

References sgpp::base::DataMatrix::get(), and python.statsfileInfo::i.

Referenced by likelihood(), setScales(), updateGP(), and var().

Member Data Documentation

std::vector<BOConfig> sgpp::datadriven::BayesianOptimization::allConfigs
protected

existing sample points in the Gaussian Process

Referenced by acquisitionOuter(), BayesianOptimization(), likelihood(), main(), setScales(), and updateGP().

double sgpp::datadriven::BayesianOptimization::bestsofar
protected

best score encountered so far (used for expected improvement calculation)

Referenced by acquisitionOuter(), BayesianOptimization(), and updateGP().

bool sgpp::datadriven::BayesianOptimization::decomFailed = false
protected

debugging variable for numerical instabilities

Referenced by decomposeCholesky(), and updateGP().

base::DataMatrix sgpp::datadriven::BayesianOptimization::gleft
protected

Cholesky Decomposition of the Gram matrix.

Referenced by setScales(), updateGP(), and var().

base::DataMatrix sgpp::datadriven::BayesianOptimization::kernelmatrix
protected

Gram matrix containing all kernel values between all existing samples.

Referenced by setScales(), updateGP(), and var().

double sgpp::datadriven::BayesianOptimization::maxofmax
protected

debugging variable for numerical instabilities

Referenced by updateGP(), and var().

base::DataVector sgpp::datadriven::BayesianOptimization::rawScores
protected

score values belonging to all existing sample points

Referenced by BayesianOptimization(), likelihood(), setScales(), and updateGP().

base::DataVector sgpp::datadriven::BayesianOptimization::scales
protected

scales of the hyperparameter space used to compute the kernel (set by fitScales())

Referenced by acquisitionOuter(), BayesianOptimization(), fitScales(), main(), setScales(), and updateGP().

bool sgpp::datadriven::BayesianOptimization::screwedvar
protected

debugging variable for numerical instabilities

Referenced by updateGP(), and var().

base::DataVector sgpp::datadriven::BayesianOptimization::transformedOutput
protected

solution of Kx = s where K is the Gram matrix and s the scores of the samples

Referenced by mean(), setScales(), and updateGP().


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