[
https://issues.apache.org/jira/browse/SPARK-7064?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14516141#comment-14516141
]
Susu Xie commented on SPARK-7064:
---------------------------------
Here is some context. During our working with Spark, we found that we can
further optimize the performance beyond the general sparse vector inside Spark.
A sparse vector inside Spark is stored as <index> + <value>. However, our
sparse vector is binary, i.e. the value is either 1 or 0. So we can eliminate
the <value> part. Another optimization is that the multiplication of values can
be optimized to sum.
> Adding binary sparse vector support
> -----------------------------------
>
> Key: SPARK-7064
> URL: https://issues.apache.org/jira/browse/SPARK-7064
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Reporter: Julien Pierre
> Priority: Minor
>
--
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
