[mlpack] Discussion about a new Idea

Ryan Curtin gth671b at mail.gatech.edu
Wed Apr 24 12:22:59 EDT 2013


On Wed, Apr 24, 2013 at 02:50:18AM +0530, ABHISHEK GUPTA wrote:
> Hi,
> I just want to suggest a new idea to incorporate some Fuzzy based methods
> to currently implemented algorithms to make them perform better in case of
> noisy or when the data is more inter-linked for example implementing Fuzzy
> C-means or fuzzy naive based classifiers. This would be really helpful for
> image based classification methods.
> I just want to know whether mlpack needs this or not.

Hello Abhishek,

I am not very familiar with the type of algorithms you are referring to.
However, that does not mean that they wouldn't be a good project.  The
only real requirement for mlpack methods is that they are faster
(preferably orders of magnitude) than competing implementations.
Generally this is done by implementing state-of-the-art algorithms that
scale better than standard approaches.

Does this help?  If not, feel free to ask more questions.

Thanks,

Ryan

-- 
Ryan Curtin       | "None of your mailman friends can hear you."
ryan at igglybob.com |   - Alpha


More information about the mlpack mailing list