[
https://issues.apache.org/jira/browse/HAMA-651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13471242#comment-13471242
]
Tommaso Teofili commented on HAMA-651:
--------------------------------------
+1, that's what I've in mind as well. I need to refactor the current
implementation a bit in order to do that since currently the total number of
inputs is not known in advance while the cost function (which usually needs
such an information) has to be calculated per each item.
I'll work in the next weeks on this and some tests / benchmarks.
> Create a gradient descent BSP
> -----------------------------
>
> Key: HAMA-651
> URL: https://issues.apache.org/jira/browse/HAMA-651
> Project: Hama
> Issue Type: New Feature
> Components: machine learning
> Affects Versions: 0.5.0
> Reporter: Tommaso Teofili
> Priority: Minor
> Fix For: 0.6.0
>
>
> A gradient descent (http://en.wikipedia.org/wiki/Gradient_descent) BSP based
> algorithm should be implemented.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
