[
https://issues.apache.org/jira/browse/SPARK-6137?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14345635#comment-14345635
]
Joseph K. Bradley commented on SPARK-6137:
------------------------------------------
Yeah, I'm not aware of a theoretical motivation for the splitting used in
Streaming K-Means.
GMeans sounds valuable to me, and the results in the paper look nice. I do
wonder if the accuracy difference with X-means in the paper's results is more
an issue with tuning parameters; it would be interesting to see results testing
whether one algorithm's hyper-parameter (alpha for GMeans vs. weighting the
model complexity penality in X-Means) was easier to tune than the other, though
that might require a whole lot of testing.
> G-Means clustering algorithm implementation
> -------------------------------------------
>
> Key: SPARK-6137
> URL: https://issues.apache.org/jira/browse/SPARK-6137
> Project: Spark
> Issue Type: New Feature
> Components: MLlib
> Reporter: Denis Dus
> Priority: Minor
>
> Will it be useful to implement G-Means clustering algorithm based on K-Means?
> G-means is a powerful extension of k-means, which uses test of cluster data
> normality to decide if it necessary to split current cluster into new two.
> It's relative complexity (compared to k-Means) is O(K), where K is maximum
> number of clusters.
> The original paper is by Greg Hamerly and Charles Elkan from University of
> California:
> [http://papers.nips.cc/paper/2526-learning-the-k-in-k-means.pdf]
> I also have a small prototype of this algorithm written in R (if anyone is
> interested in it).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
