knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ > Class Template Reference

Perform an approximate nearest neighbor search with Annoy. More...

#include <knncolle_annoy.hpp>

Inheritance diagram for knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ >:

Public Member Functions
	AnnoyBuilder (AnnoyOptions options)
	AnnoyBuilder ()=default
AnnoyOptions &	get_options ()
knncolle::Prebuilt< typename Matrix_::dimension_type, typename Matrix_::index_type, Float_ > *	build_raw (const Matrix_ &data) const
Public Member Functions inherited from knncolle::Builder< class Matrix_, typename Float_ >
std::shared_ptr< Prebuilt< typename Matrix_::dimension_type, typename Matrix_::index_type, Float_ > >	build_shared (const Matrix_ &data) const
std::unique_ptr< Prebuilt< typename Matrix_::dimension_type, typename Matrix_::index_type, Float_ > >	build_unique (const Matrix_ &data) const

Detailed Description

template<class Distance_ = Annoy::Euclidean, class Matrix_ = knncolle::SimpleMatrix<int, int, double>, typename Float_ = double, typename InternalIndex_ = typename Matrix_::index_type, typename InternalData_ = float>
class knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ >

Perform an approximate nearest neighbor search with Annoy.

In the Approximate Nearest Neighbors Oh Yeah (Annoy) algorithm, a tree is constructed where a random hyperplane splits the points into two subsets at each internal node. Leaf nodes are defined when the number of points in a subset falls below a threshold (close to twice the number of dimensions for the settings used here). Multiple trees are constructed in this manner, each of which is different due to the random choice of hyperplanes. For a given query point, each tree is searched to identify the subset of all points in the same leaf node as the query point. The union of these subsets across all trees is exhaustively searched to identify the actual nearest neighbors to the query.

See also

Bernhardsson E (2018). Annoy. https://github.com/spotify/annoy

Template Parameters

Distance_	An Annoy class to compute the distance between vectors, e.g., Annoy::Euclidean, Annoy::Manhattan. Note that this is not the same as knncolle::MockDistance.
Matrix_	Matrix-like object satisfying the knncolle::MockMatrix interface.
Float_	Floating point type for the query data and output distances.
InternalIndex_	Integer type for the internal indices in Annoy.
InternalData_	Floating point type for the internal data in Annoy. This defaults to a float instead of a double to sacrifice some accuracy for performance.

Constructor & Destructor Documentation

AnnoyBuilder() [1/2]

template<class Distance_ = Annoy::Euclidean, class Matrix_ = knncolle::SimpleMatrix<int, int, double>, typename Float_ = double, typename InternalIndex_ = typename Matrix_::index_type, typename InternalData_ = float>

knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ >::AnnoyBuilder ( AnnoyOptions  options )

inline

Parameters

options

Further options for Annoy index construction and searching.

AnnoyBuilder() [2/2]

template<class Distance_ = Annoy::Euclidean, class Matrix_ = knncolle::SimpleMatrix<int, int, double>, typename Float_ = double, typename InternalIndex_ = typename Matrix_::index_type, typename InternalData_ = float>

knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ >::AnnoyBuilder ( )

default

Default constructor.

Member Function Documentation

build_raw()

template<class Distance_ = Annoy::Euclidean, class Matrix_ = knncolle::SimpleMatrix<int, int, double>, typename Float_ = double, typename InternalIndex_ = typename Matrix_::index_type, typename InternalData_ = float>

knncolle::Prebuilt< typename Matrix_::dimension_type, typename Matrix_::index_type, Float_ > * knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ >::build_raw ( const Matrix_ &  data ) const

inlinevirtual

Creates an AnnoyPrebuilt instance.

Implements knncolle::Builder< class Matrix_, typename Float_ >.

get_options()

template<class Distance_ = Annoy::Euclidean, class Matrix_ = knncolle::SimpleMatrix<int, int, double>, typename Float_ = double, typename InternalIndex_ = typename Matrix_::index_type, typename InternalData_ = float>

AnnoyOptions & knncolle_annoy::AnnoyBuilder< Distance_, Matrix_, Float_, InternalIndex_, InternalData_ >::get_options ( )

inline

Returns

Options to the Annoy algorithm, to be modified prior to calling build_raw() and friends.

---

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

• /github/workspace/include/knncolle_annoy/knncolle_annoy.hpp