[ 
https://issues.apache.org/jira/browse/MATH-1372?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15310370#comment-15310370
 ] 

Artem Barger edited comment on MATH-1372 at 6/1/16 2:26 PM:
------------------------------------------------------------

Well, I'm using these functions currently in my work quite often, which is not 
related directly to kmeans implementations rather I'm working on some 
classification algorithms where I used them quite often. Haven't found them in 
CM, hence asked in ML whenever they needed.
 
And yes, I see myself using it while implementing "Scalable Kmeans" alg. 
described here: http://theory.stanford.edu/~sergei/papers/vldb12-kmpar.pdf. And 
for this I think decoupling of seeding procedure and actual LLoyd's iterations 
might be a good start also in term of parallelization of clustering algorithms. 


was (Author: c0rwin):
Well, I'm using these functions currently in my work quite often, which is not 
related to kmeans implementations I'm working on rather some classification 
algorithms I'm working with. Haven't found them in CM, hence asked in ML 
whenever they needed.
 
And yes, I see myself using it while implementing "Scalable Kmeans" alg. 
described here: http://theory.stanford.edu/~sergei/papers/vldb12-kmpar.pdf. And 
for this I think decoupling of seeding procedure and actual LLoyd's iterations 
might be a good start also in term of parallelization of clustering algorithms. 

> Add argmin/argmax static functions to MathArrays utility class
> --------------------------------------------------------------
>
>                 Key: MATH-1372
>                 URL: https://issues.apache.org/jira/browse/MATH-1372
>             Project: Commons Math
>          Issue Type: Wish
>            Reporter: Artem Barger
>            Assignee: Artem Barger
>            Priority: Trivial
>         Attachments: MATH-1372.patch
>
>
> Following conversation in the ML thread.
> Working lately w/ CM, I've found myself using a lot functions which helps to 
> identify the index of the min/max elements in given array of doubles. 
> In ML it was pointed out that similar functionality already exists in 
> RealVector, however due to MATH-765 it's planned to be removed (deprecated), 
> moreover to use it one has to create an instance of RealVector, where for 
> simple cases it might be redundant.
> Hence I'd like to propose to add these static methods to the MathArrays 
> utility class.



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

Reply via email to