[mlpack] GSoC - Introduction and Area of Interest.

Ryan Curtin gth671b at mail.gatech.edu
Fri Feb 28 15:38:23 EST 2014


On Fri, Feb 28, 2014 at 01:50:17AM +0530, Tejas Nikumbh wrote:
> Hi MLpack,
> 
>   I am Tejas Nikumbh,  a Fourth year Undergraduate student at IIT Bombay.
> I've done courses in *Linear Algebra, Applied Linear Algebra, Data
> Structures, Advanced Computing, Machine Learning* and have a good amount of
> programming experience. I have good project experience in simulating and
> implementing *Algorithms* as well as *Data Structures. *Some relevant
> experience involves areas like the* KD-Tree*, a *Fully Connected Neural
> Network(* This was a part of a Research Project and I scored a perfect 10
> on it) and other Data Structure implementations(template based.).
> 
>    I'm pretty familiar with Github now and can push,pull, merge, fork as
> well as submit PRs. I hope that's enough.Here is a link to some of my
> template based implementations of Datastructures (check out the directory
> that this link points to).
> https://github.com/tejasnikumbh/Datastructures<https://github.com/tejasnikumbh/Datastructures/tree/master/Vector>.
> All of my code is not on github yet; I'm in the process of revamping my git
> to make it presentable.
> 
>    I was particularly excited by the Trees Data Structure implementation
> project but noticed that it had a difficulty level score of 3/10. Does this
> mean that it is low on mlpack's priority list? I would love to research on
> the project and work on it through summers therefore please let me know if
> it is something that MLpack would consider including in its GSoC projects
> list. Also, any other feedback from the mentors as to how to go about
> working, preparing etc. is really appreciated.

Hello Tejas,

No, a low difficulty does not mean it is not important.  Conversely, I
see implementation of trees as very important (however, my perspective
is biased because I do research on trees as my job...).

The best way to get started is to download and compile mlpack, and go
through the tutorials; especially the parts of them that involve the C++
interface.  That will help give you an idea of how the code is laid out
and the code standards that are used.

For the tree project, be sure to take a close look at the code in
src/mlpack/core/tree/, because those are the implementations of the
binary space tree (which is a generalization of a kd-tree) and the cover
tree.  Note the templatization and genericity of the trees.

If there are more questions that I can answer, feel free to ask.

Thanks,

Ryan

-- 
Ryan Curtin    | "He takes such wonderful pictures with
ryan at ratml.org | his paws."  - Radio man


More information about the mlpack mailing list