Github user skavulya commented on the issue:
https://github.com/apache/flink/pull/1985
Thanks @tillrohrmann!
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes
Github user skavulya commented on the issue:
https://github.com/apache/flink/pull/1985
hi @chiwanpark sorry, I hadn't checked this PR for a while. I merged the
latest master. Do you have any preference on an alternative name for
RegularizationPenalty? Would Regularizer work better
Github user skavulya commented on the issue:
https://github.com/apache/flink/pull/1985
@chiwanpark The PR is ready. Let me know if I need to do anything else.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your
Github user skavulya commented on the issue:
https://github.com/apache/flink/pull/1985
@chiwanpark Decoupling the gradient descent step is complicated for L1
regularization because we are using the proximal gradient method that applies
soft thresholding after executing the gradient
Github user skavulya commented on the issue:
https://github.com/apache/flink/pull/1985
While working on L-BFGS, I realized that I need to remove the gradient
descent step from RegularizationPenalty. I'll update the PR soon.
---
If your project is set up for it, you can reply
Github user skavulya commented on the issue:
https://github.com/apache/flink/pull/1985
Thanks @chiwanpark. I made the changes you recommended. Please let me know
if it looks ok.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub
Github user skavulya commented on the pull request:
https://github.com/apache/flink/pull/1985
@tillrohrmann @thvasilo I made the changes you recommended. Please let me
know if they look ok.
---
If your project is set up for it, you can reply to this email and have your
reply appear
Github user skavulya commented on a diff in the pull request:
https://github.com/apache/flink/pull/1985#discussion_r63562423
--- Diff:
flink-libraries/flink-ml/src/test/scala/org/apache/flink/ml/optimization/RegularizationPenaltyITSuite.scala
---
@@ -0,0 +1,65
Github user skavulya commented on a diff in the pull request:
https://github.com/apache/flink/pull/1985#discussion_r63561773
--- Diff:
flink-libraries/flink-ml/src/main/scala/org/apache/flink/ml/optimization/GradientDescent.scala
---
@@ -272,7 +272,7 @@ abstract class
Github user skavulya commented on the pull request:
https://github.com/apache/flink/pull/656#issuecomment-218651924
thanks @thvasilo! I created the PR https://github.com/apache/flink/pull/1985
---
If your project is set up for it, you can reply to this email and have your
reply
GitHub user skavulya opened a pull request:
https://github.com/apache/flink/pull/1985
[FLINK-1979] Add logistic loss, hinge loss and regularization penalties
Thanks for contributing to Apache Flink. Before you open your pull request,
please take the following check list
Github user skavulya commented on the pull request:
https://github.com/apache/flink/pull/656#issuecomment-218380988
@tillrohrmann I updated the loss functions and regularization penalties
based on your branch. I was not sure whether to update the gradient descent
algorithm to pass
Github user skavulya commented on the pull request:
https://github.com/apache/flink/pull/1882#issuecomment-211498750
@StephanEwen Adding "Assume" statements is a good idea. We could limit them
to the affected tests.
---
If your project is set up for it, you can reply to
Github user skavulya commented on the pull request:
https://github.com/apache/flink/pull/656#issuecomment-209618076
I need this pull request for work I am doing with L-BFGS. I would like to
take it over and make the recommended changes since there has not been any
activity
GitHub user skavulya opened a pull request:
https://github.com/apache/flink/pull/1882
[FLINK-2544] Add Java 8 version for building PowerMock tests to docs
Java 8 update 11 introduced a stricter bytecode verifier that leads to
failures in unit tests that use the PowerMock runner
15 matches
Mail list logo