[mlpack] Looking to contribute for GSOC 2016

Ryan Curtin ryan at ratml.org
Sun Mar 6 13:13:41 EST 2016


On Sat, Mar 05, 2016 at 09:08:25PM +0530, Murali Raghu Babu wrote:
> Hi,
> 
>         I am a final year undergraduate student of *Computer Science and
> Engineering* at *Indian Institute of Technology, Guwahati, India*.
> 
>        I am interested in contributing to the following project ideas
> mentioned on the MLPack github page:
> 
> 1. Fast k-centers Algorithm & Implementation
> 2. Approximate nearest neighbor search
> 3. Improvement of tree traversers

Hi Murali,

There has been a lot of discussion about these projects on the mailing
list.  Below are some links to posts that may be helpful.  It would
probably be useful to search the mailing list archives further, too.

k-centers:
https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000751.html

approximate nearest neighbor:
https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000765.html

tree traversers:

I can't actually find a good email to link to for this project, so I
guess this will be the one I link to in the future. :)

A good place to start is going to be to understand the abstraction
behind mlpack's dual-tree algorithms, which is described in this paper:

http://www.ratml.org/pub/pdf/2013tree.pdf

The goal of this project is to either propose a new type of traversal,
or improve an existing type of traversal.  Here is a link to a paper
where I improved an existing traversal last year:

http://www.ratml.org/pub/pdf/2015faster.pdf

Background knowledge is going to be the most important thing for this
project, so realistically you should be familiar not just with those two
papers, but with most of the papers that those papers cite.

When you prepare your proposal, should you choose to prepare a proposal
for this project, it will be important to clearly outline your proposed
improvements (or proposed new traverser), and justify why you think it
will be faster or how it will be beneficial.

>       Could you please guide me on how exactly to proceed and get started
> on these ideas. Are these ideas already implemented ? So, Do I need to
> propose a new idea ? Do I have to solve any issues right now?

You are not obligated to contribute before submitting an application,
but it is often helpful for us to be able to see your abilities.  If you
have an entirely new idea for a proposal, you are more than welcome to
submit an application for that too.

Please let me know if I can clarify anything.

Thanks,

Ryan

-- 
Ryan Curtin    | "Because for the people who knew what it was, it
ryan at ratml.org | would be attacking their brain."


More information about the mlpack mailing list