[
https://issues.apache.org/jira/browse/MADLIB-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Frank McQuillan updated MADLIB-1293:
------------------------------------
Description:
Follow on to
https://issues.apache.org/jira/browse/MADLIB-1061
which uses a basic kd-tree implementation at the leaf node level.
This JIRA is to improve upon the basic K-D tree and add backtracking or other
methods to increase accuracy. This is an approximate method that will run
faster than brute-force, ideally for #dims up to 20-30.
was:
Follow on to
https://issues.apache.org/jira/browse/MADLIB-927
which uses brute force.
Determine other k-NN algos to implement. From
http://scikit-learn.org/stable/modules/neighbors.html
candidates are:
* K-D Tree
* Ball Tree
* Other?
This JIRA is to implement K-D tree.
> Additional computation methods for k-NN - kd tree v2
> ----------------------------------------------------
>
> Key: MADLIB-1293
> URL: https://issues.apache.org/jira/browse/MADLIB-1293
> Project: Apache MADlib
> Issue Type: New Feature
> Components: k-NN
> Reporter: Frank McQuillan
> Assignee: Orhan Kislal
> Priority: Major
> Labels: starter
> Fix For: v1.16
>
>
> Follow on to
> https://issues.apache.org/jira/browse/MADLIB-1061
> which uses a basic kd-tree implementation at the leaf node level.
> This JIRA is to improve upon the basic K-D tree and add backtracking or other
> methods to increase accuracy. This is an approximate method that will run
> faster than brute-force, ideally for #dims up to 20-30.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
