You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@spark.apache.org by "Joseph K. Bradley (JIRA)" <ji...@apache.org> on 2015/08/02 06:30:04 UTC

[jira] [Commented] (SPARK-6346) Use faster converging optimization method in MLlib

    [ https://issues.apache.org/jira/browse/SPARK-6346?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14650596#comment-14650596 ] 

Joseph K. Bradley commented on SPARK-6346:
------------------------------------------

Unsetting target version from 1.5.  We can set it to 1.6 if needed.

> Use faster converging optimization method in MLlib
> --------------------------------------------------
>
>                 Key: SPARK-6346
>                 URL: https://issues.apache.org/jira/browse/SPARK-6346
>             Project: Spark
>          Issue Type: Improvement
>          Components: ML, MLlib
>            Reporter: Reza Zadeh
>
> According to experiments in SPARK-1503, the LBFGS algorithm converges much faster than our current proximal gradient, which is used throughout MLlib. This ticket is to track replacing slower-converging algorithms, with faster components e.g. LBFGS
> This needs unification of the Optimization interface. For example, the LBFGS implementation should not know about RDDs.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org