You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by "Bae, Jae Hyeon" <me...@gmail.com> on 2011/03/20 06:59:38 UTC
Question about distributed density based clustering algorithm
Hi
I am considering to publish the implementation of density based clustering
algorithm using MapReduce.
If you have seen a similar implementation or publication, could you let me
know in detail?
I can find several papers regarding distributed density based clustering
algorithm but all of these were handling 3-D data like astronomical data.
With low dimensional data, we can use spatial index such as R* tree or
KD-tree, but my goal is implementation for very high dimensional data, text
document without dimensional reduction.
Actually, density based clustering have several limits to implement using
MapReduce. The goal is overcoming its defects efficiently to the best.
I appreciate so much if you give any kinds of comments.
Best, Jay
Re: Question about distributed density based clustering algorithm
Posted by "Bae, Jae Hyeon" <me...@gmail.com>.
Thank you so much for your reply.
By the way, I could not understand why Dirichlet process is very close to
density based clustering. Dirichlet process is representing document cluster
as a mixture model but density based clustering is finding friends of
friends using density-connectivity relationship.
Could you explain why it is close to density based clustering?
I appreciate :)
Best, Jay
2011/3/20 Ted Dunning <te...@gmail.com>
> Take a look at the Dirichlet process clustering in Mahout. I think it is
> very close to what you want.
>
>
> On Sat, Mar 19, 2011 at 10:59 PM, Bae, Jae Hyeon <me...@gmail.com>wrote:
>
>> Hi
>>
>> I am considering to publish the implementation of density based clustering
>> algorithm using MapReduce.
>>
>> If you have seen a similar implementation or publication, could you let me
>> know in detail?
>>
>> I can find several papers regarding distributed density based clustering
>> algorithm but all of these were handling 3-D data like astronomical data.
>> With low dimensional data, we can use spatial index such as R* tree or
>> KD-tree, but my goal is implementation for very high dimensional data,
>> text
>> document without dimensional reduction.
>>
>> Actually, density based clustering have several limits to implement using
>> MapReduce. The goal is overcoming its defects efficiently to the best.
>>
>> I appreciate so much if you give any kinds of comments.
>>
>> Best, Jay
>>
>
>
Re: Question about distributed density based clustering algorithm
Posted by Ted Dunning <te...@gmail.com>.
Take a look at the Dirichlet process clustering in Mahout. I think it is
very close to what you want.
On Sat, Mar 19, 2011 at 10:59 PM, Bae, Jae Hyeon <me...@gmail.com> wrote:
> Hi
>
> I am considering to publish the implementation of density based clustering
> algorithm using MapReduce.
>
> If you have seen a similar implementation or publication, could you let me
> know in detail?
>
> I can find several papers regarding distributed density based clustering
> algorithm but all of these were handling 3-D data like astronomical data.
> With low dimensional data, we can use spatial index such as R* tree or
> KD-tree, but my goal is implementation for very high dimensional data, text
> document without dimensional reduction.
>
> Actually, density based clustering have several limits to implement using
> MapReduce. The goal is overcoming its defects efficiently to the best.
>
> I appreciate so much if you give any kinds of comments.
>
> Best, Jay
>