Thanks Ryan for the quick reply.<div><br></div><div>Since I am done with the documentation part, I am now going to focus on the literature part of various trees.</div><div><br></div><div>Would it be beneficial for me to solve bugs related to project with respect to GSOC or should I concentrate on literature part?</div><div><br></div><div>Also, do you have any priorities for the projects? </div><div><br></div><div>Thanks</div><div><br></div><div>Parijat Dewangan</div><div><br></div>On Tuesday, March 8, 2016, Ryan Curtin <<a href="mailto:ryan@ratml.org">ryan@ratml.org</a>> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Mon, Mar 07, 2016 at 01:41:03PM +0530, Parijat Dewangan wrote:<br>
> Hello Marcus and Ryan,<br>
><br>
> I am working on the first idea - Implement tree types. I went through the<br>
> archive mails, followed the instructions provided there.<br>
><br>
> Following are the things I worked on till now -<br>
> 1. Building of MLPack.<br>
> 2. Going through the library of MLPack and using the already implemented<br>
> algorithms through command line.<br>
> 3. Having a sound knowledge of the paper "Tree-Independent Dual-Tree<br>
> Algorithms". Since I have a good command over data structures and<br>
> algorithms, I understood the concept of space trees and the way it is used<br>
> in implementing algorithms like k-nearest neighbors, range search etc as<br>
> explained in the paper. I have qualified for various algorithmic<br>
> competitions so understanding various trees and its implementation won't be<br>
> a problem as I have used many trees algorithms before.<br>
><br>
> Currently, I am going through the codes of the already implemented tree<br>
> types in src/mlpack/core/tree/ as the new algorithms would be based on the<br>
> same template and style. Could you specify any tree type I should start<br>
> reading about?<br>
<br>
Hi Parijat,<br>
<br>
There are lots of possible tree types you could implement. Vantage<br>
point trees (also known as metric trees) might be a good place to start,<br>
but it's certainly not the only possibility. I don't have a particular<br>
preference as to the type of tree that you pick (as long as it's not<br>
already implemented), so feel free to do your own literature search and<br>
find the tree you are most interested in.<br>
<br>
> While I am very interested in the previously mentioned project, I am also<br>
> eager to work on Decision trees' and 'Fast k-centers algorithms and its<br>
> implementation'.<br>
<br>
Here are some mailing list links (in case you haven't already seen<br>
them), but be aware that each of these projects have significant amounts<br>
of background material to comprehend and it may therefore be useful to<br>
focus on only one project.<br>
<br>
<a href="https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000751.html" target="_blank">https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000751.html</a><br>
<a href="https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000796.html" target="_blank">https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/000796.html</a><br>
<br>
> Is there any work/reading that you think I should do in order to be better<br>
> prepared to write a good proposal for GSOC 2016.<br>
<br>
There's the application guide, and you can also read a useful student<br>
manual too:<br>
<br>
<a href="https://github.com/mlpack/mlpack/wiki/Google-Summer-of-Code-Application-Guide" target="_blank">https://github.com/mlpack/mlpack/wiki/Google-Summer-of-Code-Application-Guide</a><br>
<a href="http://write.flossmanuals.net/gsocstudentguide/what-is-google-summer-of-code/" target="_blank">http://write.flossmanuals.net/gsocstudentguide/what-is-google-summer-of-code/</a><br>
<br>
> I appreciate the documentation of MLPack. It is well written and<br>
> systematic. :)<br>
<br>
Thanks! I am glad you found the documentation useful.<br>
<br>
Ryan<br>
<br>
--<br>
Ryan Curtin | "You got to stick with your principles."<br>
<a href="javascript:;" onclick="_e(event, 'cvml', 'ryan@ratml.org')">ryan@ratml.org</a> | - Harry Waters<br>
</blockquote>