File Clustering.h

namespace faiss

Copyright (c) Facebook, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree.

Throughout the library, vectors are provided as float * pointers. Most algorithms can be optimized when several vectors are processed (added/searched) together in a batch. In this case, they are passed in as a matrix. When n vectors of size d are provided as float * x, component j of vector i is

x[ i * d + j ]

where 0 <= i < n and 0 <= j < d. In other words, matrices are always compact. When specifying the size of the matrix, we call it an n*d matrix, which implies a row-major storage.

Copyright (c) Facebook, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. I/O functions can read/write to a filename, a file handle or to an object that abstracts the medium.

The read functions return objects that should be deallocated with delete. All references within these objectes are owned by the object.

Copyright (c) Facebook, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. Definition of inverted lists + a few common classes that implement the interface.

Copyright (c) Facebook, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. Since IVF (inverted file) indexes are of so much use for large-scale use cases, we group a few functions related to them in this small library. Most functions work both on IndexIVFs and IndexIVFs embedded within an IndexPreTransform.

Copyright (c) Facebook, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. In this file are the implementations of extra metrics beyond L2 and inner product

Copyright (c) Facebook, Inc. and its affiliates.

This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. Defines a few objects that apply transformations to a set of vectors Often these are pre-processing steps.

Functions

float kmeans_clustering(size_t d, size_t n, size_t k, const float *x, float *centroids)

simplified interface

Parameters

d – dimension of the data
n – nb of training vectors
k – nb of output centroids
x – training set (size n * d)
centroids – output centroids (size k * d)

Returns

final quantization error

struct Clustering : public faiss::ClusteringParameters 

#include <Clustering.h>

K-means clustering based on assignment - centroid update iterations

The clustering is based on an Index object that assigns training points to the centroids. Therefore, at each iteration the centroids are added to the index.

On output, the centoids table is set to the latest version of the centroids and they are also added to the index. If the centroids table it is not empty on input, it is also used for initialization.

Subclassed by faiss::Clustering1D

Public Types

typedef Index::idx_t idx_t

Public Functions

Clustering(int d, int k)

Clustering(int d, int k, const ClusteringParameters &cp)

virtual void train(idx_t n, const float *x, faiss::Index &index, const float *x_weights = nullptr)

run k-means training

Parameters

x – training vectors, size n * d
index – index used for assignment
x_weights – weight associated to each vector: NULL or size n

void train_encoded(idx_t nx, const uint8_t *x_in, const Index *codec, Index &index, const float *weights = nullptr)

run with encoded vectors

win addition to train()’s parameters takes a codec as parameter to decode the input vectors.

Parameters: codec – codec used to decode the vectors (nullptr = vectors are in fact floats) *

void post_process_centroids(): Post-process the centroids after each centroid update. includes optional L2 normalization and nearest integer rounding

inline virtual ~Clustering()

Public Members

size_t d: dimension of the vectors

size_t k: nb of centroids

std::vector<float> centroids: centroids (k * d) if centroids are set on input to train, they will be used as initialization

std::vector<ClusteringIterationStats> iteration_stats: stats at every iteration of clustering

int niter: clustering iterations

int nredo: redo clustering this many times and keep best

bool verbose

bool spherical: do we want normalized centroids?

bool int_centroids: round centroids coordinates to integer

bool update_index: re-train index after each iteration?

bool frozen_centroids: use the centroids provided as input and do not change them during iterations

int min_points_per_centroid: otherwise you get a warning

int max_points_per_centroid: to limit size of dataset

int seed: seed for the random number generator

size_t decode_block_size: how many vectors at a time to decode

struct Clustering1D : public faiss::Clustering 

#include <Clustering.h>

Exact 1D clustering algorithm

Since it does not use an index, it does not overload the train() function

Public Types

typedef Index::idx_t idx_t

Public Functions

explicit Clustering1D(int k)

Clustering1D(int k, const ClusteringParameters &cp)

void train_exact(idx_t n, const float *x)

inline virtual ~Clustering1D()

virtual void train(idx_t n, const float *x, faiss::Index &index, const float *x_weights = nullptr)

run k-means training

Parameters

x – training vectors, size n * d
index – index used for assignment
x_weights – weight associated to each vector: NULL or size n

void train_encoded(idx_t nx, const uint8_t *x_in, const Index *codec, Index &index, const float *weights = nullptr)

run with encoded vectors

win addition to train()’s parameters takes a codec as parameter to decode the input vectors.

Parameters: codec – codec used to decode the vectors (nullptr = vectors are in fact floats) *

void post_process_centroids(): Post-process the centroids after each centroid update. includes optional L2 normalization and nearest integer rounding

Public Members

size_t d: dimension of the vectors

size_t k: nb of centroids

std::vector<float> centroids: centroids (k * d) if centroids are set on input to train, they will be used as initialization

std::vector<ClusteringIterationStats> iteration_stats: stats at every iteration of clustering

int niter: clustering iterations

int nredo: redo clustering this many times and keep best

bool verbose

bool spherical: do we want normalized centroids?

bool int_centroids: round centroids coordinates to integer

bool update_index: re-train index after each iteration?

bool frozen_centroids: use the centroids provided as input and do not change them during iterations

int min_points_per_centroid: otherwise you get a warning

int max_points_per_centroid: to limit size of dataset

int seed: seed for the random number generator

size_t decode_block_size: how many vectors at a time to decode

struct ClusteringIterationStats

Public Members

float obj: objective values (sum of distances reported by index)

double time: seconds for iteration

double time_search: seconds for just search

double imbalance_factor: imbalance factor of iteration

int nsplit: number of cluster splits

struct ClusteringParameters

#include <Clustering.h>

Class for the clustering parameters. Can be passed to the constructor of the Clustering object.

Subclassed by faiss::Clustering, faiss::ProgressiveDimClusteringParameters

Public Functions

ClusteringParameters(): sets reasonable defaults

Public Members

int niter: clustering iterations

int nredo: redo clustering this many times and keep best

bool verbose

bool spherical: do we want normalized centroids?

bool int_centroids: round centroids coordinates to integer

bool update_index: re-train index after each iteration?

bool frozen_centroids: use the centroids provided as input and do not change them during iterations

int min_points_per_centroid: otherwise you get a warning

int max_points_per_centroid: to limit size of dataset

int seed: seed for the random number generator

size_t decode_block_size: how many vectors at a time to decode

struct ProgressiveDimClustering : public faiss::ProgressiveDimClusteringParameters 

#include <Clustering.h>

K-means clustering with progressive dimensions used

The clustering first happens in dim 1, then with exponentially increasing dimension until d (I steps). This is typically applied after a PCA transformation (optional). Reference:

“Improved Residual Vector Quantization for High-dimensional Approximate

Nearest Neighbor Search”

Shicong Liu, Hongtao Lu, Junru Shao, AAAI’15

https://arxiv.org/abs/1509.05195

Public Types

using idx_t = Index::idx_t 

Public Functions

ProgressiveDimClustering(int d, int k)

ProgressiveDimClustering(int d, int k, const ProgressiveDimClusteringParameters &cp)

void train(idx_t n, const float *x, ProgressiveDimIndexFactory &factory)

inline virtual ~ProgressiveDimClustering()

Public Members

size_t d: dimension of the vectors

size_t k: nb of centroids

std::vector<float> centroids: centroids (k * d)

std::vector<ClusteringIterationStats> iteration_stats: stats at every iteration of clustering

int progressive_dim_steps: number of incremental steps

bool apply_pca: apply PCA on input

int niter: clustering iterations

int nredo: redo clustering this many times and keep best

bool verbose

bool spherical: do we want normalized centroids?

bool int_centroids: round centroids coordinates to integer

bool update_index: re-train index after each iteration?

bool frozen_centroids: use the centroids provided as input and do not change them during iterations

int min_points_per_centroid: otherwise you get a warning

int max_points_per_centroid: to limit size of dataset

int seed: seed for the random number generator

size_t decode_block_size: how many vectors at a time to decode

struct ProgressiveDimClusteringParameters : public faiss::ClusteringParameters 

Subclassed by faiss::ProgressiveDimClustering

Public Functions

ProgressiveDimClusteringParameters()

Public Members

int progressive_dim_steps: number of incremental steps

bool apply_pca: apply PCA on input

int niter: clustering iterations

int nredo: redo clustering this many times and keep best

bool verbose

bool spherical: do we want normalized centroids?

bool int_centroids: round centroids coordinates to integer

bool update_index: re-train index after each iteration?

bool frozen_centroids: use the centroids provided as input and do not change them during iterations

int min_points_per_centroid: otherwise you get a warning

int max_points_per_centroid: to limit size of dataset

int seed: seed for the random number generator

size_t decode_block_size: how many vectors at a time to decode

struct ProgressiveDimIndexFactory

#include <Clustering.h>

generates an index suitable for clustering when called

Subclassed by faiss::gpu::GpuProgressiveDimIndexFactory

Public Functions

virtual Index *operator()(int dim): ownership transferred to caller

inline virtual ~ProgressiveDimIndexFactory()