[
https://issues.apache.org/jira/browse/MADLIB-1272?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16596671#comment-16596671
]
Rahul Iyer commented on MADLIB-1272:
------------------------------------
[PR|https://github.com/apache/madlib/pull/313] created for fixing this issue.
> Simplify Nesterov by performing velocity update in previous iteration
> ---------------------------------------------------------------------
>
> Key: MADLIB-1272
> URL: https://issues.apache.org/jira/browse/MADLIB-1272
> Project: Apache MADlib
> Issue Type: Bug
> Components: Module: Neural Networks
> Reporter: Rahul Iyer
> Priority: Major
>
> Momentum updates are complicated due to Nesterov requiring an initial update
> before gradient calculations. There is, however, a different form of the
> Nesterov update that can be cleanly performed after the regular update,
> simplifying the code. This allows performing the gradient calculations before
> any update - with or without Nesterov.
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
