[ https://issues.apache.org/jira/browse/FLINK-1994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15013142#comment-15013142 ]
Till Rohrmann commented on FLINK-1994: -------------------------------------- Hi Trevor, would be great to add tests and docs, otherwise it happens easily that someone else changes things without noticing that affects also other parts. What you can do though is to open the PR and then adding commits with the documentation and tests to the same branch. That way we can already review the gain implementation. Cheers, Till > 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)