[
https://issues.apache.org/jira/browse/SINGA-313?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16022786#comment-16022786
]
ASF subversion and git services commented on SINGA-313:
-------------------------------------------------------
Commit 0815391d0488acbe3f788d71457c28ec4b083dee in incubator-singa's branch
refs/heads/master from wangwei
[ https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git;h=0815391 ]
SINGA-313 Add L2 norm layer
> Add L2 norm layer
> -----------------
>
> Key: SINGA-313
> URL: https://issues.apache.org/jira/browse/SINGA-313
> Project: Singa
> Issue Type: New Feature
> Reporter: wangwei
>
> L2 norm layer is useful for triplet ranking loss, which normalizes the inputs
> within the same scale (the L2 norm is 1) and thus avoids numerical problems
> (e.g. exploding).
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
