[
https://issues.apache.org/jira/browse/SPARK-1503?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14595226#comment-14595226
]
Joseph K. Bradley commented on SPARK-1503:
------------------------------------------
[~staple] [~lewuathe] Can you please coordinate about how convergence is
measured, as in [SPARK-3382]? The current implementation for [SPARK-3382] uses
relative convergence w.r.t. the weight vector, where it measures relative to
the weight vector from the previous iteration. I figure we should use the same
convergence criterion for both accelerated and non-accelerated gradient descent.
> Implement Nesterov's accelerated first-order method
> ---------------------------------------------------
>
> Key: SPARK-1503
> URL: https://issues.apache.org/jira/browse/SPARK-1503
> Project: Spark
> Issue Type: New Feature
> Components: MLlib
> Reporter: Xiangrui Meng
> Assignee: Aaron Staple
> Attachments: linear.png, linear_l1.png, logistic.png, logistic_l2.png
>
>
> Nesterov's accelerated first-order method is a drop-in replacement for
> steepest descent but it converges much faster. We should implement this
> method and compare its performance with existing algorithms, including SGD
> and L-BFGS.
> TFOCS (http://cvxr.com/tfocs/) is a reference implementation of Nesterov's
> method and its variants on composite objectives.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
