You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@mahout.apache.org by "Angel Martinez Gonzalez (JIRA)" <ji...@apache.org> on 2013/05/07 15:31:15 UTC

[jira] [Updated] (MAHOUT-775) L2 does not work with TrainAdaptiveLogisticRegression

     [ https://issues.apache.org/jira/browse/MAHOUT-775?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Angel Martinez Gonzalez updated MAHOUT-775:
-------------------------------------------

    Attachment: MAHOUT-775.patch

The problem is that the L2 variance ("s" member)is initialized with a 0 value. As regularization is controlled by AdaptiveLogisticRegression through the "lambda" hyperparameter, I think a value of 1 would be ok for "s". Also, a variance of 0 does never make sense, so I modified the constructor of L2.
See attached patch.
                
> L2 does not work with TrainAdaptiveLogisticRegression
> -----------------------------------------------------
>
>                 Key: MAHOUT-775
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-775
>             Project: Mahout
>          Issue Type: Bug
>          Components: Classification
>    Affects Versions: 0.6
>            Reporter: XiaoboGu
>         Attachments: MAHOUT-775.patch
>
>
> I have post the problem to the dev list, see the following message
> http://mail-archives.apache.org/mod_mbox/mahout-dev/201106.mbox/%3cBANLkTik6153pjgCFNAyuPrWbV9jzCXPPJA@mail.gmail.com%3e

--
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