You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Oleg Ignatenko (JIRA)" <ji...@apache.org> on 2017/09/24 12:52:00 UTC

[jira] [Comment Edited] (IGNITE-6059) Use any distributed matrix in K-Means

    [ https://issues.apache.org/jira/browse/IGNITE-6059?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16178182#comment-16178182 ] 

Oleg Ignatenko edited comment on IGNITE-6059 at 9/24/17 12:51 PM:
------------------------------------------------------------------

[~mlipkovich] the answer to your question is _no_ and, _maybe_, see below.

_No_ - because [current interface of this clusterer|https://github.com/apache/ignite/blob/master/modules/ml/src/main/java/org/apache/ignite/ml/clustering/KMeansDistributedClusterer.java| see it at Ignite master on Github] explicitly requires sparse distributed matrix:

 {code}KMeansDistributedClusterer extends BaseKMeansClusterer<SparseDistributedMatrix>{code}

_Maybe_ - because when I gave it a quick try and changed class and method signatures to use AbstractMatrix instead of SparseDistributedMatrix compiler didn't complain.

A way to ensure if this replacement is legitimate that comes to mind is to replicate currently existing unit tests and examples involving this clusterer to similar code that uses some other implementation of AbstractMatrix and verify that these pass.


was (Author: oignatenko):
[~mlipkovich] the answer to your question is _no_ and, _maybe_, see below.

_No_ - because current interface of this clusterer explicitly requires sparse distributed matrix:

 {code}KMeansDistributedClusterer extends BaseKMeansClusterer<SparseDistributedMatrix>{code}

_Maybe_ - because when I gave it a quick try and changed class and method signatures to use AbstractMatrix instead of SparseDistributedMatrix compiler didn't complain.

A way to ensure if this replacement is legitimate that comes to mind is to replicate currently existing unit tests and examples involving this clusterer to similar code that uses some other implementation of AbstractMatrix and verify that these pass.

> Use any distributed matrix in K-Means
> -------------------------------------
>
>                 Key: IGNITE-6059
>                 URL: https://issues.apache.org/jira/browse/IGNITE-6059
>             Project: Ignite
>          Issue Type: Improvement
>          Components: ml
>            Reporter: Yury Babak
>             Fix For: 2.3
>
>
> Currently k-means work only with row/col matrix.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)