Data structure for Nearest Neighbor Search - VP Trees

Posted by orp on Tuesday January 31, 2012 @10:14AM
from the pnylab.com dept.

http://pnylab.com/pny/papers/vptree/vptree/

( 3 comments | 5 points | on HN )

Re:Data structure for Nearest Neighbor Search - VP Trees by arnoldoMuller on Wednesday February 01, 2012 @10:57PM (#3541351)

This is a very important data structure from a historical point of view, but unfortunately it doesn't work well when you are dealing with complex spaces. Try LSH as mentioned before or something that is 10X faster: http://simmachines.com

[ Parent | on HN ]

Re:Data structure for Nearest Neighbor Search - VP Trees by pork on Tuesday January 31, 2012 @12:42PM (#3534276)

Don't waste your time on space partitioning trees if you have big data, jump straight to Locality Sensitive Hashing or approximate nearest neighbor graph construction (WWW 2011)

[ Parent | on HN ]

Re:Data structure for Nearest Neighbor Search - VP Trees by BigZaphod on Tuesday January 31, 2012 @11:41AM (#3533989)

I'm no expert, but this doesn't seem to be any significantly better than a k-d tree, does it? I might be missing something. I find academic papers hard to parse.

[ Parent | on HN ]