spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Joseph K. Bradley (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SPARK-7770) Should GBT validationTol be relative tolerance?
Date Fri, 11 Sep 2015 20:39:46 GMT

    [ https://issues.apache.org/jira/browse/SPARK-7770?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14741505#comment-14741505
] 

Joseph K. Bradley commented on SPARK-7770:
------------------------------------------

I had intended for us to discuss this some.  Thinking more about it, perhaps we should mimic
convergenceTol in GradientDescent.  That switches between relative and absolute error, depending
on the magnitude of the error.  What do you think?

> Should GBT validationTol be relative tolerance?
> -----------------------------------------------
>
>                 Key: SPARK-7770
>                 URL: https://issues.apache.org/jira/browse/SPARK-7770
>             Project: Spark
>          Issue Type: Improvement
>          Components: ML, MLlib
>            Reporter: Joseph K. Bradley
>            Assignee: Yanbo Liang
>            Priority: Minor
>
> In spark.mllib, GBT validationTol uses absolute tolerance.  Relative tolerance is arguably
easier to set in a meaningful way.  Questions:
> * Should we change spark.mllib's validationTol meaning?
> * Should we use relative tolerance in spark.ml's GBT (once we add validation support)?
> I would vote for changing both to relative tolerance, where the tolerance is relative
to the current loss on the training set.



--
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


Mime
View raw message