| Title: | Fast and Scalable Approximate k-Nearest Neighbor Search Methods using 'N2' Library |
|---|---|
| Description: | Implements methods to perform fast approximate K-nearest neighbor search on input matrix. Algorithm based on the 'N2' implementation of an approximate nearest neighbor search using hierarchical Navigable Small World (NSW) graphs. The original algorithm is described in "Efficient and Robust Approximate Nearest Neighbor Search Using Hierarchical Navigable Small World Graphs", Y. Malkov and D. Yashunin, <doi:10.1109/TPAMI.2018.2889473>, <arXiv:1603.09320>. |
| Authors: | Peter Kharchenko [aut], Viktor Petukhov [aut], Dirk Eddelbuettel [ctb], Evan Biederstedt [cre, aut] |
| Maintainer: | Evan Biederstedt <[email protected]> |
| License: | Apache License 2.0 |
| Version: | 1.0.3 |
| Built: | 2024-11-22 04:37:03 UTC |
| Source: | https://github.com/kharchenkolab/n2r |
boolean to check OpenMP exists
checkOpenMP()checkOpenMP()
Perform fast approximate K-nearest neighbor search of rows input matrix mA in rows of matrix mB.
crossKnn( mA, mB, k, nThreads = 10L, verbose = TRUE, indexType = "angular", M = 12L, MaxM0 = 24L, ef_search_multiplier = 50, quiet = FALSE )crossKnn( mA, mB, k, nThreads = 10L, verbose = TRUE, indexType = "angular", M = 12L, MaxM0 = 24L, ef_search_multiplier = 50, quiet = FALSE )
mA |
Input numeric matrix of data |
mB |
Input numeric matrix of data |
k |
Integer number of clusters |
nThreads |
Integer number of threads (default=10) |
verbose |
Boolean flag for verbose output (default=FALSE) |
indexType |
Metric distance type, which can be "angular" or "L2" (default="angular") |
M |
Integer number of connections (default=12) The NSW graph is constructed via consecutive insertion of elements in random order by bidirectionally connecting them to the M closest neighbors from the previously inserted elements. |
MaxM0 |
Integer maximum number of connections that an element can have in the zero layer. (default=24) It is recommended that MaxM0 not exceed 2*M. |
ef_search_multiplier |
Integer multiplier to calculate candidate nearest neighbors, set to k*ef_search_multiplier (default=50). Refer to the parameters er and efConstruction in Malkov & Yashunin (2020) doi: 10.1109/TPAMI.2018.2889473 |
quiet |
Boolean flag specifically for Rcpp warnings (default=FALSE) |
clusters per row in sparse Matrix of class "dgCMatrix" of dimensions mB rows by mA rows
data(iris) iris_df = data.matrix(iris[-5]) ## convert to a numeric matrix crossKnn(mA=iris_df, mB=head(iris_df, 50), 4)data(iris) iris_df = data.matrix(iris[-5]) ## convert to a numeric matrix crossKnn(mA=iris_df, mB=head(iris_df, 50), 4)
Perform fast approximate K-nearest neighbor search on rows of the input matrix m.
Knn( m, k, nThreads = 10L, verbose = TRUE, indexType = "angular", M = 12L, MaxM0 = 24L, ef_search_multiplier = 50, quiet = FALSE )Knn( m, k, nThreads = 10L, verbose = TRUE, indexType = "angular", M = 12L, MaxM0 = 24L, ef_search_multiplier = 50, quiet = FALSE )
m |
Input numeric matrix of data |
k |
Integer number of clusters |
nThreads |
Integer number of threads (default=10) |
verbose |
Boolean flag for verbose output (default=FALSE) |
indexType |
Metric distance type, which can be "angular" or "L2" (default="angular") |
M |
Integer number of connections (default=12) The NSW graph is constructed via consecutive insertion of elements in random order by bidirectionally connecting them to the M closest neighbors from the previously inserted elements. |
MaxM0 |
Integer maximum number of connections that an element can have in the zero layer. (default=24) It is recommended that MaxM0 not exceed 2*M. |
ef_search_multiplier |
Integer multiplier to calculate candidate nearest neighbors, set to k*ef_search_multiplier (default=50). Refer to the parameters er and efConstruction in Malkov & Yashunin (2020) doi: 10.1109/TPAMI.2018.2889473 |
quiet |
Boolean flag specifically for Rcpp warnings (default=FALSE) |
clusters per row in sparse Matrix of class "dgCMatrix" of dimensions m rows by m rows
data(iris) iris_df = data.matrix(iris[-5]) ## convert to a numeric matrix Knn(m=iris_df, 4)data(iris) iris_df = data.matrix(iris[-5]) ## convert to a numeric matrix Knn(m=iris_df, 4)