hama-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hudson (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HAMA-668) Fix possible number instabilities in LogisticRegression CF
Date Wed, 21 Nov 2012 10:07:58 GMT

    [ https://issues.apache.org/jira/browse/HAMA-668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13501823#comment-13501823
] 

Hudson commented on HAMA-668:
-----------------------------

Integrated in Hama trunk #60 (See [https://builds.apache.org/job/Hama%20trunk/60/])
    [HAMA-668] - let LogisticRegressionModel use BigDecimals internally to avoid number instabilities
(Revision 1412031)

     Result = SUCCESS
tommaso : 
Files : 
* /hama/trunk/examples/src/test/java/org/apache/hama/examples/GradientDescentTest.java
* /hama/trunk/examples/src/test/java/org/apache/hama/examples/LinearRegressionTest.java
* /hama/trunk/examples/src/test/java/org/apache/hama/examples/LogisticRegressionTest.java
* /hama/trunk/examples/src/test/resources/gd_file_sample.txt
* /hama/trunk/examples/src/test/resources/linear_regression_sample.txt
* /hama/trunk/examples/src/test/resources/logistic_regression_sample.txt
* /hama/trunk/ml/src/main/java/org/apache/hama/ml/regression/LogisticRegressionModel.java
* /hama/trunk/ml/src/test/java/org/apache/hama/ml/regression/LogisticRegressionModelTest.java

                
> Fix possible number instabilities in LogisticRegression CF
> ----------------------------------------------------------
>
>                 Key: HAMA-668
>                 URL: https://issues.apache.org/jira/browse/HAMA-668
>             Project: Hama
>          Issue Type: Bug
>          Components: machine learning
>            Reporter: Tommaso Teofili
>            Assignee: Tommaso Teofili
>             Fix For: 0.6.0
>
>
> In calculateCostForItem the log of several calculations may run in the special cases
of the logarithm:
> * If the argument is NaN or less than zero, then the result is NaN.
> * If the argument is positive infinity, then the result is positive infinity.
> * If the argument is positive zero or negative zero, then the result is negative infinity.
> We should guard them.

--
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

Mime
View raw message