[jira] [Commented] (SINGA-174) Add Batch Normalization layer and Local Response Nomalization layer.

["ASF subversion and git services (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SINGA-174?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15313799#comment-15313799 ] 

ASF subversion and git services commented on SINGA-174:
-------------------------------------------------------

Commit fa2ea304d8989818a80780c9f428e0fcc19db031 in incubator-singa's branch refs/heads/dev from [~flytosky]
[ https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git;h=fa2ea30 ]

SINGA-174 Add Batch Normalization layer and Local Response Normalization layer.

Remove buffering input/output if the flag of Layer::Forward() is not kTrain.
The input/output are used during Layer::Backward() which only appears
for kTrain.


> Add Batch Normalization layer and Local Response Nomalization layer.
> --------------------------------------------------------------------
>
>                 Key: SINGA-174
>                 URL: https://issues.apache.org/jira/browse/SINGA-174
>             Project: Singa
>          Issue Type: New Feature
>         Environment: All
>            Reporter: Wang Ji
>            Assignee: Wang Ji
>
> The Batch Normalization layer and LRN layer are implemented using Tensor math functions.
> CudnnBatchNorm and CudnnLRN layer are implemented using cudnn 5.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)