Finding K Nearest neighbors: ~30% data access, >1000 dimensional uniform spaces.

Werner Van Belle¹ - werner@yellowcouch.org, werner.van.belle@gmail.com

1- Yellowcouch;

Abstract : This talk describes a method I developed to find the k nearest neighbors in large dimensional spaces with less than 30% data access. The research was conducted between 2010 and 2013 but only 2 days before the slides were made I figured out that the "VA-Files" approach (invented in 1998 by Weber) is very similar.

Keywords: k-NN search, k nearest neighbors search, multi dimensional queries
Reference: Werner Van Belle; Finding K Nearest neighbors: ~30% data access, >1000 dimensional uniform spaces.; Presented at Observe, Hack, Make 2013: Oudkarpsel; Netherlands; August 2013
See also:
The (preliminary) article describing the development of the layered bit approach

http://werner.yellowcouch.org/
werner@yellowcouch.org