Jump to content

Talk:Vantage-point tree

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Proposed merge with MVP tree

[edit]

MVP tree is a generalization of the better-known VP tree. It can be adequately covered in the more complete Vantage-point tree article. Pburka (talk) 22:29, 16 November 2014 (UTC)[reply]

 Done JackPotte (talk) 00:06, 30 January 2016 (UTC)[reply]

I added notes about Brin's GNAT data structure, which is related to the MVP. Brin's paper: http://www.vldb.org/conf/1995/P574.PDF --Ericjster (talk) 00:59, 29 January 2019 (UTC)[reply]

Desired New Sections

[edit]

Algorithms are missing: a construction algorithm as well as a nearest neighbor search algorithms would be welcome. Also, a search for all neighbors within a tolerance margin would be welcome. 160.129.138.225 (talk) 15:04, 23 March 2017 (UTC)[reply]

Each variant has some differences in the construction, and those details often can be important in practice. The construction algorithm should generalize the different variations. --Ericjster (talk) 00:59, 29 January 2019 (UTC)[reply]

I think more should be said about metric spaces, and all you need is a function to compare two things. For example one can create an index on text or pictures, and the user only needs to provide a distance function. The distance function must maintain the triangle inequality. --Ericjster (talk) 00:59, 29 January 2019 (UTC)[reply]