[
https://issues.apache.org/jira/browse/FLINK-1994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15088548#comment-15088548
]
ASF GitHub Bot commented on FLINK-1994:
---------------------------------------
Github user chiwanpark commented on the pull request:
https://github.com/apache/flink/pull/1397#issuecomment-169861378
Hi @rawkintrevo, I'm sorry about waiting you.
I have looked your pull request. Almost of changes are good but I have some
few comments.
First, there are some meaningless changes such as indentation in
`flink-ml/pom.xml` file and definition of `optimize` method of
`GradientDescent` class. Please revert this changes.
Second, how about using `Enum` for the added parameter? Numbers cannot
express meaning of optimization method.
Third, I think that we should apply this changes to documentation. Please
add this content into the FlinkML documentation. :)
> Add different gain calculation schemes to SGD
> ---------------------------------------------
>
> Key: FLINK-1994
> URL: https://issues.apache.org/jira/browse/FLINK-1994
> Project: Flink
> Issue Type: Improvement
> Components: Machine Learning Library
> Reporter: Till Rohrmann
> Assignee: Trevor Grant
> Priority: Minor
> Labels: ML, Starter
>
> The current SGD implementation uses as gain for the weight updates the
> formula {{stepsize/sqrt(iterationNumber)}}. It would be good to make the gain
> calculation configurable and to provide different strategies for that. For
> example:
> * stepsize/(1 + iterationNumber)
> * stepsize*(1 + regularization * stepsize * iterationNumber)^(-3/4)
> See also how to properly select the gains [1].
> Resources:
> [1] http://arxiv.org/pdf/1107.2490.pdf
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
