[mlpack] [GSOC-2016] Prospective Applicant interested in working in mlpack
Ryan Curtin
ryan at ratml.org
Wed Mar 2 11:30:06 EST 2016
On Wed, Mar 02, 2016 at 01:02:22PM +0300, Sritanu Chakraborty wrote:
> Let me take this opportunity to introduce myself.
>
> *Brief Introduction:*
> I am Sritanu Chakraborty from India. I am pursuing my Masters in Advanced
> Combinatorics(2015-2017) at Moscow Institute of Physics and
> Technology,Moscow,Russia.
>
> ...
>
> *Choice of Projects(*in order of preference*):*
> Well,to be very honest I am spoilt for choice(s).All of the project
> mentioned are extremely interesting.However,though I have indepth knowledge
> about data structures,algorithms and programming ,I don't have extensive
> background in machine learning and neural networks.The following three
> projects I have chosen in order of preference:
> 1.Implement Tree Types
> 2.Fast k-centers Algorithm and Implementation.
> 3.Approximate Nearest Neighbor Search
>
> At the time of the project I will be in Kolkata,India(End of May-August).
> Looking forward to your reply.
Hi Sritanu,
Good to meet you.
Lots has been written on each of those three projects, so I'd suggest
you take a look through the mailing list archives to see what you can
find:
https://mailman.cc.gatech.edu/pipermail/mlpack/
(Maybe you can search that with Google, that might make it easier.
Mailman doesn't provide a nice search interface or anything.)
Here's some information on the tree types project that I just posted:
https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000760.html
Here's some information on the fast k-centers project that I just
posted:
https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000751.html
As for the approximate nearest neighbor search project, a good place to
start would actually be the same place as for the other two projects:
familiarize yourself with tree-based algorithms, read papers on
dual-tree algorithms, and also read papers on approximate nearest
neighbor search techniques.
Once you've done that, you could probably spend some time looking
through the mlpack nearest neighbor search code in
src/mlpack/methods/neighbor_search/ to see how the current exact nearest
neighbor search code works, and think about how to extend that to the
approximate case.
I hope this is helpful; please let me know if I can clarify anything.
Thanks!
Ryan
--
Ryan Curtin | "And the last thing I would ever do is lie to you."
ryan at ratml.org | - Marlon
More information about the mlpack
mailing list