[mlpack] Automative benchmarking

[Ranjani]

Hi,
Thanks for the link on Shogun's integration.I looked at their detailed
expectations.
I would like to discuss more on
1.Adding new algorithms
2.Improving accuracy and fixing Bottlenecks


As visualization part using d3js is for representing results I would like
to discuss on them a little later.


Thanks
Shree Ranjani


On Tue, Mar 4, 2014 at 9:23 PM, Marcus Edel <marcus.edel at fu-berlin.de>wrote:

> Hello Shree.
>
> Great that you're interested in working on the automatic benchmarking
> project!
>
> 1.Given a specific time frame ,which is the fastest running time.
> 2.Benchmarking in distributed environment.
> Also,I feel that the benchmarking scripts can also be extended to
> classification algorithms like SVM,Bayesian as well.
>
>
> We can definitely expand the benchmark system in this direction. The
> tricky part is to visualize everything in a good manner. I like the
> suggestion from the shogun description to use d3js (take a look at the
> examples):
>
> https://github.com/mbostock/d3/wiki/Gallery
>
> It would be really nice if you could give a clear idea about the
> integration with Shogun in this project.
>
>
> As pointed out in their project description, the main task will be focused
> on implementing a way to compare the accuracy of the algorithms (like you
> suggested), adding more shogun algorithms to the comparison and fix
> potential bottlenecks. I think that's just a one line summary of their
> description:
>
> http://www.shogun-toolbox.org/page/Events/gsoc2014_ideas#mlpack
>
> Anyway, maybe you can tell us which part you like to discuss in more
> detail.
>
> Thanks,
> Marcus
>
>
> On 04 Mar 2014, at 14:08, Ranjani <shreeranjaniss at gmail.com> wrote:
>
> Hi,
> I am Shree Ranjani ,a final year Computer Science student from India.
>
> I am really interested in continuing the work on "*Automative
> benchmarking of mlpack methods*".
> I have been playing around with mlpack libraries quite a bit and also ran
> benchmarking scripts for some algorithms.I have been going through the
> discussions on this project and find that the following ideas should be
> given more priority more than others
>
> 1.Given a specific time frame ,which is the fastest running time.
> 2.Benchmarking in distributed environment.
> Also,I feel that the benchmarking scripts can also be extended to
> classification algorithms like SVM,Bayesian as well.
>
> It would be really nice if you could give a clear idea about the
> integration with Shogun in this project.
>
> Mentioning more about myself - I am pretty comfortable writing codes in
> Python,C++. I have done a research internship in NetApp,India on a machine
> learning project.
> I am really interested in *Machine learning ,Data Mining and NLP* and all
> my academic projects were in this area.Some of them are
> 1.Developed an Intelligent Classroom system that automated subjective
> question and answer evaluation and provided feedback to teachers about
> students' performance
> 2.Working on Strength of Sentiment analysis of movie data using NLP
> techniques
> I have taken up courses on Artificial Intelligence,Data Mining and Machine
> learning during my course curriculum.
>
>
> Do let me know,so that i can start working asap! :)
>
> --
> Regards,
> Shree Ranjani
>
>  _______________________________________________
> mlpack mailing list
> mlpack at cc.gatech.edu
> https://mailman.cc.gatech.edu/mailman/listinfo/mlpack
>
>
>


-- 
Regards,
Shree Ranjani
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cc.gatech.edu/pipermail/mlpack/attachments/20140305/22190763/attachment-0003.html>