07:31 < barun_> Hello community

11:15 < KeonKim> hello

11:16 < KeonKim> I am interested in participating in Google Summer of Code :>

11:16 < KeonKim> especially in <Dataset and Experimentation Tools> or <Decision Trees>

11:40 < toshad> rcurtin: I was working on issue #275 (cover tree memory implementation). I have read the cover tree algorithm and was trying to understand the code. I couldn't understand what exactly the empty statistic class does.

11:43 < zoq> KeonKim: Hello, you are free to send an application for any of the projects.

13:17 < rcurtin> toshad: the statistic type is a template parameter, and depending on the algorithm used, a different statistic may be used

13:24 < toshad> so, the statistic just stores some extra information regarding the points and in case of empty statistic, its stoes nothing. Right?

13:24 < rcurtin> correct

13:26 < toshad> Do you have any method in mind as to how the nodes should be arranged??

13:26 < toshad> I couldn't think of anything other than a simple adjacency vector for every node

14:28 < Keon> so basically, <Dataset and experimentation tools> is a data preprocessing tool right?

14:29 < zoq> Keon: right

14:33 < Keon> I just wanted to hear 1. functionalities you guys already built 2. the functions you guys are using the most with python libraries such as pandas 3. the functions that are not in already existing tools but you guys think should be included

14:34 < Keon> Lastly, any other resources I can read to improve my understandings?

14:42 < zoq> Keon: You could take a look at the archive and maybe find more information on the project there.: https://mailman.cc.gatech.edu/pipermail/mlpack/2016-March/thread.html

16:08 < praveench> Hi Marcus, I had a look at the tests suggested by you the other day

16:08 < praveench> Found them quite interesting, will be fun to implement them

16:09 < praveench> Yesterday I had written down the implementation of an RBM as explicitly as possible

16:11 < praveench> I found it to be not that long. I think it would be properly done in two weeks time. And the next two weeks for writing the tests

16:12 < praveench> so in a month RBM is done. Similarly for DBN, as its based on RBM hence another month for that

16:13 < praveench> I will have technically got another month to go

16:13 < praveench> Am I correct in thinking so

16:14 < praveench> I am quite new to open source field even though I have extensively written codes

16:14 < praveench> Github @chvsp

16:15 < praveench> I am quite comfortable with the ann implemented till date

16:26 < wasiq> rcurtin, hello I submitted my proposal yesterday, and it would be great if you review it when you have time.

16:51 < cloudh> hmm, hello there, I have some question about the GSoC 2016 application. On https://github.com/mlpack/mlpack/wiki/SummerOfCodeIdeas it says "These are a list of ideas compiled by mlpack developers", however it also says "At the bottom are a list of projects from 2013 and 2014 that aren't really applicable to this year, or were done.", so I'm confused about what does the list mean.

16:53 < cloudh> Is the list ranged from "Implement tree types" to "Decision trees" an applicable list or non-applicable one?

16:56 < zoq> cloudh: Hello, you are free to choose any idea from "Implement tree types" to "Decision trees". Every project under 'Old ideas for GSoC', aren't really applicable to this year.

16:58 < cloudh> ok thx. 1 am here, gotta sleep

18:11 < ranjan123> I want to discuss about the implementation of scd. can any one help me about this ?

18:27 < daydayup> Hello everyone, I am applying for GSoC and have a questions. I am doing my PhD actually but I read on the GitHub page that GSoC is like a full time job. My PhD research is related to some of the mlpack projects, especially the trees data structures, KNN, Kmeans, Kcenters, etc. I wonder if your group just accepts applicants who will work on your projects, or you also support and allow people to work on their own new but related idea

19:10 < rcurtin> daydayup: we are definitely open to other ideas too

19:11 < rcurtin> still the amount of work should be equivalent to a full time job for the summer

19:11 < rcurtin> if it overlaps with your research that's okay as long as there is a coding component to it (this is what Google cares about) and it should fit in well with the rest of mlpack

19:12 < rcurtin> Google may have other restrictions, I am not sure; you will have to read their information

19:24 < ranjan123> Hello rcurtin: you there ?

22:37 < uzipaz> rcurtin: I wish to use SGD to optimize a linear model on a given dataset, how do I specify the class that implements Evaluate, Gradient functions to be used by SGD implementation...