[
https://issues.apache.org/jira/browse/SPARK-5016?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14320935#comment-14320935
]
Xiangrui Meng commented on SPARK-5016:
--------------------------------------
I think we should compute the inverse in parallel. In
https://github.com/apache/spark/blob/master/mllib/src/main/scala/org/apache/spark/mllib/clustering/GaussianMixture.scala#L166,
we don't collect to local by use aggregateByKey to save the sums to reducers.
Then on each reducer, we update the Guassians, and finally collect them to the
driver.
> GaussianMixtureEM should distribute matrix inverse for large numFeatures, k
> ---------------------------------------------------------------------------
>
> Key: SPARK-5016
> URL: https://issues.apache.org/jira/browse/SPARK-5016
> Project: Spark
> Issue Type: Improvement
> Components: MLlib
> Affects Versions: 1.2.0
> Reporter: Joseph K. Bradley
>
> If numFeatures or k are large, GMM EM should distribute the matrix inverse
> computation for Gaussian initialization.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
