hama-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Behroz Sikander (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HAMA-941) Semiclustering Termination
Date Sat, 30 Apr 2016 14:09:13 GMT

    [ https://issues.apache.org/jira/browse/HAMA-941?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15265326#comment-15265326
] 

Behroz Sikander commented on HAMA-941:
--------------------------------------

In the past few days, I went through the material on semi-clustering. I have developed basic
understanding of the algorithm but there is no real example implementation available online.


Then I studied the code of semi-clustering implemented in Hama. I do not think that semi-clustering
algorithm is working properly. The score of a semi-cluster remains 1.0 and it never changes
(this 1.0 value is defined in the constructor of SemiClusterDetails class). Further, the most
strange thing is that the code (semiClusterScoreCalcuation) to actually calculate the Score
never fires !.

In the "compute" event of "SemiClusteringVertex" class, the following condition is never satisfied
and since the condition is never satisfied, the score is never calculated.

if (!msg.contains(this.getVertexID())
            && msg.size() == semiClusterMaximumVertexCount) {
.....
msgNew.setScore(semiClusterScoreCalcuation(msgNew));
.....
}

How should we proceed ? I can look into the Giraph implementation of Semi-Clustering and can
try to find out what is the problem with semi-clustering. (https://github.com/grafos-ml/okapi/blob/master/src/main/java/ml/grafos/okapi/graphs/SemiClustering.java)

> Semiclustering Termination
> --------------------------
>
>                 Key: HAMA-941
>                 URL: https://issues.apache.org/jira/browse/HAMA-941
>             Project: Hama
>          Issue Type: Improvement
>          Components: examples, graph
>            Reporter: Edward J. Yoon
>            Priority: Minor
>
> Currently Semiclustering example will be terminated when the number of iterations exceeded
the predefined threshold max iteration.
> App should be stopped if there's no cluster changes (I guess). Please check and improve
it.



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

Mime
View raw message