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 descent step. I left the
regularization penalty as-is. I am thinking of adding an additional method that
adds the regularization penalty to gradient without the gradient descent step
but I will do it in the L-BFGS PR instead.
---
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 so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
