And for machine learning, I just heard Prof. Ng on the
ml-class.orgspeaking about neural networks.

In my opinion Neural Networks could be implemented, trained and queried
very optimal with BSP, because there is a huge amount of communication
between the layers.
Currently I don't have a straight idea, but I will put this on my TODO list
for the next few months.

2011/11/2 Thomas Jungblut <[email protected]>

> Clustering: There was a k-means in BSP on github. [1]
>
> I wanted to blog about my own implementation, but I don't want to work at
> the I/O stuff once again.
> Once I fixed SSSP and Pagerank, I can have a look at the k-means.
>
> [1]
> https://github.com/willmore/hama-kmeans/blob/master/src/main/java/ee/ut/cs/willmore/KMeansCluster.java
>
>
> 2011/11/2 Tommaso Teofili <[email protected]>
>
>> 2011/11/2 Edward J. Yoon <[email protected]>
>>
>> > Hi,
>> >
>> > I'm plan to move unuseful and low-quality examples to test package.
>> > Then, only few examples (Pi, SSSP, PageRank) are left.
>> >
>> > It'd be nice If we can add some (clearly differentiable from other
>> > solutions) ML/Mining/Clustering example.
>> >
>> > Do you have any good idea?
>> >
>>
>> I'd be very interested in that part (for example enable running Mahout or
>> UIMA on top of Hama) but I'm lacking the needed time right now. But if
>> someone would start I may be able to help.
>> Tommaso
>>
>>
>>
>> >
>> > --
>> > Best Regards, Edward J. Yoon
>> > @eddieyoon
>> >
>>
>
>
>
> --
> Thomas Jungblut
> Berlin <[email protected]>
>



-- 
Thomas Jungblut
Berlin <[email protected]>

Reply via email to