Ubuntu
python-pynndescent package

python-pynndescent 0.5.2+dfsg-1 source package in Ubuntu

Changelog

python-pynndescent (0.5.2+dfsg-1) unstable; urgency=medium

  * Team Upload.
  [ Andreas Tille ]
  * New upstream version
  * Exclude *.pyc files from upstream source
  * Make package Architecture dependent to exclude 32bit architectures
    where the tests are failing
  * Remove empty boilerplate

  [ Nilesh Patra ]
  * d/rules: Enable build time tests respecting DEB_BUILD_OPTIONS
  * d/tests/control: Restrict testing architectures to supported
    build architectures
  * d/salsa-ci.yml: Do run build on i386

 -- Nilesh Patra <email address hidden>  Sat, 03 Jul 2021 23:45:46 +0530

Upload details

Uploaded by:
Debian Med
Uploaded to:
Sid
Original maintainer:
Debian Med
Architectures:
any-amd64 arm64 mips64el ppc64el s390x ia64 ppc64 riscv64 sparc64 alpha
Section:
misc
Urgency:
Medium Urgency

See full publishing history Publishing

Series Pocket Published Component Section
Jammy release universe misc

Downloads

File Size SHA-256 Checksum
python-pynndescent_0.5.2+dfsg-1.dsc 2.4 KiB 38f4205869167fd1f644b7728c5fad7a6098d756d4a6667f008c8771a78d7435
python-pynndescent_0.5.2+dfsg.orig.tar.xz 934.8 KiB 9d9608281140b4fcd521b07c3636c12aac77e2125e70ed144e11a7cf077910b6
python-pynndescent_0.5.2+dfsg-1.debian.tar.xz 3.5 KiB 417fa5f29cd16da0514c45d4fa2085f08e63aa7724949023c7e56c52ba02bd3e

Available diffs

No changes file available.

Binary packages built by this source

python3-pynndescent: nearest neighbor descent for approximate nearest neighbors

 PyNNDescent is a Python nearest neighbor descent for approximate nearest
 neighbors. It provides a Python implementation of Nearest Neighbor
 Descent for k-neighbor-graph construction and approximate nearest
 neighbor search, as per the paper:
 .
 Dong, Wei, Charikar Moses, and Kai Li. "Efficient k-nearest neighbor
 graph construction for generic similarity measures." Proceedings of the
 20th international conference on World wide web. ACM, 2011.
 .
 This library supplements that approach with the use of random projection
 trees for initialisation. This can be particularly useful for the
 metrics that are amenable to such approaches (euclidean, minkowski,
 angular, cosine, etc.). Graph diversification is also performed, pruning
 the longest edges of any triangles in the graph.
 .
 Currently this library targets relatively high accuracy (80%-100%
 accuracy rate) approximate nearest neighbor searches.