You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by Grant Ingersoll <gs...@apache.org> on 2009/06/25 22:10:58 UTC
Re: [jira] Commented: (MAHOUT-121) Speed up distance calculations for sparse vectors
On Jun 25, 2009, at 3:41 PM, Ted Dunning wrote:
> THat is one option.
>
> Another is to have a separate distance measure that doesn't have the
> square
> root.
>
Yeah, but that implies that we have one interface for all the other
DistanceMeasures and one for Euclidean, or, I suppose we could have a
PseudoEuclideanDM class that doesn't do the square root from which the
Euclidean merely wraps and gives back the sqrt. Actually, I like that
solution, except the name part of it...
> On Thu, Jun 25, 2009 at 12:40 PM, Grant Ingersoll (JIRA) <jira@apache.org
> >wrote:
>
>>
>> [
>> https://issues.apache.org/jira/browse/MAHOUT-121?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12724228
>> #action_12724228]
>>
>> Grant Ingersoll commented on MAHOUT-121:
>> ----------------------------------------
>>
>> BTW, any objection to me removing the sqrt calculation from
>> EuclideanDistanceMeasure?
>>
>>
Re: [jira] Commented: (MAHOUT-121) Speed up distance calculations for sparse vectors
Posted by Grant Ingersoll <gs...@apache.org>.
On Jun 25, 2009, at 4:10 PM, Grant Ingersoll wrote:
>
> On Jun 25, 2009, at 3:41 PM, Ted Dunning wrote:
>
>> THat is one option.
>>
>> Another is to have a separate distance measure that doesn't have
>> the square
>> root.
>>
>
> Yeah, but that implies that we have one interface for all the other
> DistanceMeasures and one for Euclidean, or, I suppose we could have
> a PseudoEuclideanDM class that doesn't do the square root from which
> the Euclidean merely wraps and gives back the sqrt. Actually, I
> like that solution, except the name part of it...
>
I'm going to go with the base class approach. Seems like that should
work. Committing shortly.
Re: [jira] Commented: (MAHOUT-121) Speed up distance calculations
for sparse vectors
Posted by Jeff Eastman <jd...@windwardsolutions.com>.
+1
Ted Dunning wrote:
> That is what I meant.
>
> As for name, how about SquaredEuclideanDistance?
>
> On Thu, Jun 25, 2009 at 1:10 PM, Grant Ingersoll <gs...@apache.org>wrote:
>
>
>> I suppose we could have a PseudoEuclideanDM class that doesn't do the
>> square root from which the Euclidean merely wraps and gives back the sqrt.
>> Actually, I like that solution, except the name part of it
>>
>
>
Re: [jira] Commented: (MAHOUT-121) Speed up distance calculations for
sparse vectors
Posted by Ted Dunning <te...@gmail.com>.
That is what I meant.
As for name, how about SquaredEuclideanDistance?
On Thu, Jun 25, 2009 at 1:10 PM, Grant Ingersoll <gs...@apache.org>wrote:
> I suppose we could have a PseudoEuclideanDM class that doesn't do the
> square root from which the Euclidean merely wraps and gives back the sqrt.
> Actually, I like that solution, except the name part of it