[mlpack] Hello from a student
Ryan Curtin
gth671b at mail.gatech.edu
Mon Apr 15 10:55:44 EDT 2013
On Sun, Apr 14, 2013 at 12:25:30PM +0530, Rishi Mukherjee wrote:
> Hello,
>
> I am an undergraduate student of Computer Science Engineering from India. I
> went through the list of ideas of mlpack and found 2 of the ideas very
> interesting. The first one is "python and r bindings for mlpack". I have
> been programming since the last 2.5 years with python. Also, I was a
> speaker at PyCon India 2012. I am interested in algorithms and machine
> learning and I think while implementing this project I will learn
> state-of-art machine learning algorithms.
Hello Rishi,
During the course of that project you will be exposed to the user
interface of those machine learning algorithms more than the algorithmic
side of it. We already have the algorithms written in C++ and the goal
is to provide an interface to each of the methods to languages like R,
Python, and MATLAB. If you are more interested in writing a specific
machine learning algorithm, you can always propose one. Still, knowing
how a machine learning method is used is often more useful in practice
than knowing exactly how the algorithm works. For instance, I may know
the EM algorithm very well, but I'm not very useful if I don't know
where it can be applied. :)
> The second project I am interested in is "Collaborative filtering package".
> I have always wanted to know how websites suggest items to customers.
> Implementing a package for this will be a great source of learning about
> them, after that I may use it to write an example application using one of
> the available web frameworks.
There has been a lot of talk about the collaborative filtering package.
Here a thread you can read that has good information:
https://mailman.cc.gatech.edu/pipermail/mlpack/2013-April/000034.html
If you have other questions, feel free to ask.
> Please suggest me how a machine learning beginner like me can start
> contributing to mlpack.
A good way to get started is to check out the current development
version of the code with subversion and get acquainted with the way
things are laid out and what mlpack offers. Another good idea would be
to build mlpack and play around with the programs on toy datasets.
There are some tutorials at http://www.mlpack.org/tutorial.html which
you may find useful.
Thanks,
Ryan
--
Ryan Curtin | "Give a man a gun and he thinks he's Superman.
ryan at igglybob.com | Give him two and he thinks he's God." - Pang
More information about the mlpack
mailing list