[ https://issues.apache.org/jira/browse/FLINK-1994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15023699#comment-15023699 ]
ASF GitHub Bot commented on FLINK-1994: --------------------------------------- GitHub user rawkintrevo opened a pull request: https://github.com/apache/flink/pull/1397 [FLINK-1994] [ml] Add different gain calculation schemes to SGD Continuation of pull request 1384. Fixed long line issue, rebased on top of current master, and changed title per instructions. https://github.com/apache/flink/pull/1384 You can merge this pull request into a Git repository by running: $ git pull https://github.com/rawkintrevo/flink flink-1994 Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/1397.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #1397 ---- commit bbf452e7cac7fcafcc494c79062d5139fe49117c Author: rawkintrevo <trevor.d.gr...@gmail.com> Date: 2015-10-05T16:27:58Z FLINK-1994: Added 3 new effective learning rates commit 08c39f68cd30ce039131f2a1928fb7ae8125e1a2 Author: trevor <tre...@tower1.hsd1.il.comcast.net> Date: 2015-10-31T00:28:53Z Merge remote-tracking branch 'upstream/master' I'm not sure what's happening... commit acb37eb9e13a45bd622cf98708616d96acab4fc9 Author: Trevor Grant <trevor.d.gr...@gmail.com> Date: 2015-11-19T01:31:52Z Added SGD gain calculation schemes commit 09312af1f4e5069a7389f4fb19ebcfa5cbc07d3d Author: Trevor Grant <trevor.d.gr...@gmail.com> Date: 2015-11-19T02:02:55Z fixed optimal SGD calculation scheme commit 223150eead99149a06b0b8e32a6ea947c5c4aad3 Author: rawkintrevo <trevor.d.gr...@gmail.com> Date: 2015-10-05T16:27:58Z FLINK-1994: Added 3 new effective learning rates commit 61b7597c699cbd2fda878ca65657af3acf641537 Author: Trevor Grant <trevor.d.gr...@gmail.com> Date: 2015-11-19T01:31:52Z Added SGD gain calculation schemes commit 0cb36ecf272d2c8da8f7341c52604d1226487b1a Author: Trevor Grant <trevor.d.gr...@gmail.com> Date: 2015-11-19T02:02:55Z fixed optimal SGD calculation scheme commit 955c00c8a5e999a882a0b0da7be4c72b29638a36 Author: Trevor Grant <trevor.d.gr...@gmail.com> Date: 2015-11-24T03:09:06Z [FLINK-1994] [ml] Add different gain calculation schemes to SGD commit 2a2f77efaffcf5bd2340ef181abd35b84fc7b760 Author: Trevor Grant <trevor.d.gr...@gmail.com> Date: 2015-11-24T03:18:40Z [FLINK-1994] [ml] Add different gain calculation schemes to SGD ---- > 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)