[
https://issues.apache.org/jira/browse/MAHOUT-121?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12719462#action_12719462
]
Shashikant Kore commented on MAHOUT-121:
----------------------------------------
+1 Grant's suggestion that we split two issues.
Sean, Doubling the vector size looks aggressive. In a recent experiment, I ran
into trouble due to such aggressive expansion. We could optimize memory usage
and Sparse Vector initialization for input document vectors by exposing some
low-level APIs. When document vector is read, we already know the number of
elements in it. It need not go through the iterative set() method which avoids
the movement of the array elements. FastIntDouble can be initialized by two
parallel arrays created by reading the sparse vector formatted string.
I will try out your patch and report performance.
> Speed up distance calculations for sparse vectors
> -------------------------------------------------
>
> Key: MAHOUT-121
> URL: https://issues.apache.org/jira/browse/MAHOUT-121
> Project: Mahout
> Issue Type: Improvement
> Components: Matrix
> Reporter: Shashikant Kore
> Attachments: mahout-121.patch, Mahout1211.patch
>
>
> From my mail to the Mahout mailing list.
> I am working on clustering a dataset which has thousands of sparse vectors.
> The complete dataset has few tens of thousands of feature items but each
> vector has only couple of hundred feature items. For this, there is an
> optimization in distance calculation, a link to which I found the archives of
> Mahout mailing list.
> http://lingpipe-blog.com/2009/03/12/speeding-up-k-means-clustering-algebra-sparse-vectors/
> I tried out this optimization. The test setup had 2000 document vectors
> with few hundred items. I ran canopy generation with Euclidean distance and
> t1, t2 values as 250 and 200.
>
> Current Canopy Generation: 28 min 15 sec.
> Canopy Generation with distance optimization: 1 min 38 sec.
> I know by experience that using Integer, Double objects instead of primitives
> is computationally expensive. I changed the sparse vector implementation to
> used primitive collections by Trove [
> http://trove4j.sourceforge.net/ ].
> Distance optimization with Trove: 59 sec
> Current canopy generation with Trove: 21 min 55 sec
> To sum, these two optimizations reduced cluster generation time by a 97%.
> Currently, I have made the changes for Euclidean Distance, Canopy and KMeans.
>
> Licensing of Trove seems to be an issue which needs to be addressed.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
