[
https://issues.apache.org/jira/browse/SPARK-19208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15822549#comment-15822549
]
Ilya Matiach commented on SPARK-19208:
--------------------------------------
[~srowen] isn't feature hashing (eg HashingTF) to large bit sizes a common
scenario (and then using SVMs or linear learners) for large text data - eg 2^20
bit sizes. In addition, even if there are just a few columns but a ton of data
we are wasting a lot of performance by computing other metrics. This seems like
a good change, but maybe it would instead be better to modify the api to the
summarizer to only provide a subset of statistics - that way you would both 1.)
not duplicate the code 2.) offer more flexibility to users. Otherwise, I am
fine with the code change, but it would be good to see more statistics on how
the performance has improved, since this is a significant code change and we
need to be convinced that it is a positive change (no perf regressions) for all
types of data.
> MaxAbsScaler and MinMaxScaler are very inefficient
> --------------------------------------------------
>
> Key: SPARK-19208
> URL: https://issues.apache.org/jira/browse/SPARK-19208
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Reporter: zhengruifeng
> Assignee: Apache Spark
> Attachments: WechatIMG2621.jpeg
>
>
> Now, {{MaxAbsScaler}} and {{MinMaxScaler}} are using
> {{MultivariateOnlineSummarizer}} to compute the min/max.
> However {{MultivariateOnlineSummarizer}} will also compute extra unused
> statistics. It slows down the task, moreover it is more prone to cause OOM.
> For example:
> env : --driver-memory 4G --executor-memory 1G --num-executors 4
> data:
> [http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#kdd2010%20(bridge%20to%20algebra)]
> 748401 instances, and 29,890,095 features
> {{MaxAbsScaler.fit}} fails because of OOM
> {{MultivariateOnlineSummarizer}} maintains 8 arrays:
> {code}
> private var currMean: Array[Double] = _
> private var currM2n: Array[Double] = _
> private var currM2: Array[Double] = _
> private var currL1: Array[Double] = _
> private var totalCnt: Long = 0
> private var totalWeightSum: Double = 0.0
> private var weightSquareSum: Double = 0.0
> private var weightSum: Array[Double] = _
> private var nnz: Array[Long] = _
> private var currMax: Array[Double] = _
> private var currMin: Array[Double] = _
> {code}
> For {{MaxAbsScaler}}, only 1 array is needed (max of abs value)
> For {{MinMaxScaler}}, only 3 arrays are needed (max, min, nnz)
> After modication in the pr, the above example run successfully.
--
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
