SketchHBE Class Reference

#include <SketchHBE.h>

Inheritance diagram for SketchHBE:

Public Member Functions
	SketchHBE (shared_ptr< MatrixXd > X, int M, double w, int k, shared_ptr< Kernel > ker)
	SketchHBE (shared_ptr< MatrixXd > X, int M, double w, int k, int scales, shared_ptr< Kernel > ker)
	SketchHBE (shared_ptr< MatrixXd > X, vector< SketchTable > &sketches, vector< vector< int >> &indices, int M, double w, int k, shared_ptr< Kernel > ker, std::mt19937_64 &rng)
Public Member Functions inherited from MoMEstimator
double	query (VectorXd q, double lb, int m)

Public Attributes
vector< HashTable >	tables
Public Attributes inherited from MoMEstimator
double	totalTime = 0

Protected Member Functions
double	evaluateQuery (VectorXd query)
vector< double >	MoM (VectorXd query, int L, int m)

Detailed Description

HBE on HBS

Constructor & Destructor Documentation

SketchHBE() [1/3]

SketchHBE::SketchHBE	(	shared_ptr< MatrixXd >	X,
		int	M,
		double	w,
		int	k,
		shared_ptr< Kernel >	ker
	)

Build N_SKETCHES number of sketches (HBS); sample from sketches to build hash tables for HBE

Parameters

X	full dataset
M	number of samples
w	bin width
k	number of hash functions
ker	kernel function

SketchHBE() [2/3]

SketchHBE::SketchHBE	(	shared_ptr< MatrixXd >	X,
		int	M,
		double	w,
		int	k,
		int	scales,
		shared_ptr< Kernel >	ker
	)

Build N_SKETCHES number of sketches (HBS); sample from sketches to build hash tables for HBE

Parameters

X	full dataset
M	number of samples
w	bin width
k	number of hash functions
scales	number of weight scales in hash buckets
ker	kernel function

SketchHBE() [3/3]

SketchHBE::SketchHBE	(	shared_ptr< MatrixXd >	X,
		vector< SketchTable > &	sketches,
		vector< vector< int >> &	indices,
		int	M,
		double	w,
		int	k,
		shared_ptr< Kernel >	ker,
		std::mt19937_64 &	rng
	)

Build HBE from points sampled from given sketches

Parameters

X	full dataset
sketches	HBS to sample points from
indices	maps a sample (index) from HBS to an index in the whole dataset
M	number of samples
w	bin width
k	number of hash functions
ker	kernel function
rng	random number generator

Member Function Documentation

evaluateQuery()

double SketchHBE::evaluateQuery ( VectorXd  query )

protected

Take a biased sample from a hash table via HBE.

Parameters

query

query point

Returns

normalized contribution of the biased sample

MoM()

vector< double > SketchHBE::MoM ( VectorXd  q, int  L, int  m  )

protectedvirtual

Parameters

q	query
L	median of L means
m	means of m samples

Returns

: a vector of L elements, where each element is a sum of m samples

Implements MoMEstimator.

Member Data Documentation

tables

vector<HashTable> SketchHBE::tables

A collection of hash tables for HBE.

---

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

• SketchHBE.h
• SketchHBE.cpp