09:30 < kartik98> Hi, I am new to mlpack. I get this error when i run mlpack_test: Library not loaded. https://imgur.com/gallery/iZ1zn . Can someone help me out?

09:31 < kartik98> I am building mlpack on mac

11:38 < hodor123456> I am feeling TMP very hard.Please somebody help...

12:45 < rcurtin> I'm not sure what TMP is

12:54 < zoq> I guess, template meta programming

12:55 < zoq> There are quite a few links on http://www.mlpack.org/gsoc.html e.g. http://www.drdobbs.com/policy-based-design-in-the-real-world/184401861 about policy-based design.

13:31 < hodor123456> Thanks a lot,but they are not enough.I really want to contribute on mlpack but feeling numb on metaprogramming.Please do help.

13:40 < rcurtin> hodor123456: you could start from those tutorials and try and search for more information that might help

13:40 < rcurtin> or possibly find some C++ book that introduces templates and start from there

13:41 < zoq> I really like Andrei Alexandrescu's book Modern C++ Design.

14:02 < hodor123456> May I know on which educational field and year you people are.I am feeling very dumb.

14:11 < hodor123456> Actually I am in total dilemma.Should I stick with Mlpack or move on(for GSoC-2018) with another languages.I have about 1.5 month before orgs list come out.Should I full focus on TMP(14 hours a day) or move on :( ?Please guide me.

14:15 < rcurtin> hodor123456: to answer the first question, I finished a PhD about two years ago

14:15 < rcurtin> for the second, I can't say---that is a decision for you to make

14:16 < rcurtin> but to be honest it does not sound like very much fun to study template metaprogramming so hard for months

14:16 < rcurtin> and maybe there is some other organization that your existing skillset fits with better

14:17 < rcurtin> but if you really are more interested in mlpack than other organizations, then go ahead and learn---up to you

14:17 < rcurtin> but do realize that we haven't even applied yet to be a mentoring organization, so it is possible that we may not be accepted (it has happened before)

14:20 < hodor123456> Would you suggest me some orgs base on C++.I know Python,PHP,Javascript,little JAVA.But I find C++ love.Please.

14:21 < rcurtin> I have no idea, to be honest; I don't know what kind of projects you would enjoy

14:21 < rcurtin> I would suggest looking through the list of past organizations that have been accepted

14:22 < hodor123456> Thanks a lot for your kind help :)

14:23 < rcurtin> sure, I hope that you find what you are looking for

14:24 < hodor123456> A last question,How many students got selected from mlpack & what was there educational level?

14:24 < rcurtin> if you're interested in having us help with the bugs you opened, please feel free, otherwise, you should feel free to go ahead and close them

14:24 < rcurtin> last year we accepted 10 of I think 69 or 70 applications

14:24 < rcurtin> of those students, most were third or fourth year undergraduates and a few were Master's or PhD students

14:30 < alsc> this is so cool… the t-sne visualization came for free using this TF’s tool

14:30 < alsc> http://projector.tensorflow.org/?config=https://gist.githubusercontent.com/alesaccoia/a71a3adf475bd84ec84b57842cf8814b/raw/48f932222619abcbee8c94ec60829ad5d691e35f/projector_config.json

14:31 < alsc> after fitting t-sne,the outliers are clearly shown.. bliss

14:37 < dilder_> Sorry,but I am not done with those issues.Should I close them?

14:39 < rcurtin> dilder_: are you the same person as hodor123456: if so, absolutely, if you are not done with the issues then you should not close them, we can continue to dsbug :)

14:39 < rcurtin> *debug

14:39 < rcurtin> alsc: great to hear, sounds like t-sne will make it easy for filtering

14:41 < alsc> for what I understand, when you said I could use it as embedding you meant I should get hold of the projection matrix and run KNN in this new space right?

14:47 < rcurtin> yeah, that was the idea, it might work better with KNN

14:47 < rcurtin> T-SNE could be more helpful since it's a nonlinear transform, whereas NCA is linear

14:48 < rcurtin> I don't think there's any guarantee it will do better but at least intuitively it seems to me that it should

18:25 < Rupal> Hi

18:26 < Rupal> I am new, and I would like to contribute. I have worked previously in ml and ai.

18:26 < rcurtin> Rupal: hello there

18:26 < rcurtin> you might try looking at http://www.mlpack.org/involved.htmk

18:26 < rcurtin> er, involved.html, sorry for the tpo

18:26 < rcurtin> typo* (I am not doing very well at typing today I guess)

18:26 < Rupal> yeah gotcha

18:34 < Rupal> I would like to keep working on this for quite some time as I am finding this pretty interesting. I would like to extend it to work on a project if you offer in google summer of code. How should I start? Should I start with fixing issues

18:40 < zoq> Rupal: Hello, we have a special page for GSoC: http://mlpack.org/gsoc.html that might answer some questions.

18:42 < zoq> If you find some interesting issue on GitHub feel free to work on it; also we are always interested in new interesting methods, so if you like to implement something new let's discuss.

18:43 < Rupal> Ok, thank you.

18:46 < Rupal> I am not so great at C++ is it possible for me to catch up if I put in effort and interest.

18:51 < zoq> Sure, mlpack does use a lot of template based paradigms, some are listed on the GSoC page, so I would recommend to check them out.