[jira] [Created] (SPARK-31032) GMM compute summary and update distributions in one pass

["zhengruifeng (Jira)" <ji...@apache.org>]

zhengruifeng created SPARK-31032:
------------------------------------

             Summary: GMM compute summary and update distributions in one pass
                 Key: SPARK-31032
                 URL: https://issues.apache.org/jira/browse/SPARK-31032
             Project: Spark
          Issue Type: Improvement
          Components: ML
    Affects Versions: 3.1.0
            Reporter: zhengruifeng


In current impl, GMM need to trigger two jobs at one iteration:

1, one to compute summary;

{color:#172b4d}2, if {{{color}{color:#c7a65d}{color:#172b4d}shouldDistributeGaussians}} {color}{color}((k - {color:#4dacf0}1.0{color}) / k) * numFeatures > {color:#4dacf0}25.0,{color}

{color:#c7a65d}{color:#172b4d}trigger another to update distributions;{color}
{color}

 

{color:#c7a65d}{color:#172b4d}shouldDistributeGaussians is almost true in practice, since numFeatures is likely to be greater than 25.{color}{color}

 

{color:#c7a65d}{color:#172b4d}We can use only one job to impl above computation,{color}{color}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

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