01:04 < AlanSanchezGitte> Hi, hope you all are ok and stay safe during the COVID breakout. I'm Alan Sanchez, a 4th-year student majoring in Applied Math in UASLP, México. I'm pretty new to mlpack since I've worked a bit of ML just using R. I've taken many upper-division courses; two of them are Multivariate Statistics and Time Series. I'm accustomed to Python due to Coding Theory schemes and programs, but I would like to get familiar with

01:04 < AlanSanchezGitte> mlpack package. My main programming language is C++, where I've worked in Oriented-object programming. Hope I may learn a lot from you. I'll be trying to help out and keep updated during the GSoC process!

01:06 < zoq> AlanSanchezGitte: Welcome, let us know if there is anything that needs clarification, also in case you haven't seen it yet we have a dedicated GSoC page - https://www.mlpack.org/gsoc.html that has some starting pointers.

03:19 < yfgeng6Gitter[m]> Hi everyone. I’m GENG Yunfan, a postgraduate in University of California Irvine and got my bachelor degree in computer science from Chinese University of Hong Kong. Due to the COVID-19, I am currently working as a research intern in Autonomous Driving team of an AI company, Sensetime in China and my duty is to design some decision algorithms. I have taken some ML and deep learning courses so I think I am familiar

03:19 < yfgeng6Gitter[m]> with CV, Graph embedding and RL. My intern job right now is relevant with RL and main coding language is c++. As a CS student I can code with python, C/C++. I am new to mlpack and GSoC but I think it will be a good chance for me to learn from the community whether or not I can get into this program. I am looking forward to be able to contribute to the open source communities!

03:54 < NippunSharmaGitt> Hey @yfgeng6 welcome to the community. A good way to start can be to build mlpack from source [https://www.mlpack.org/doc/mlpack-git/doxygen/build.html](https://www.mlpack.org/doc/mlpack-git/doxygen/build.html) you can also look at [https://www.mlpack.org/gsoc.html](https://www.mlpack.org/gsoc.html) for more information on GSoC.

03:58 < yfgeng6Gitter[m]> Thanks @NippunSharma I have started reading these documents 😏

04:12 < NippunSharmaGitt> 👍

06:37 < RishabhGargGitte> > `rcurtin (@ryan:ratml.org)` Basically in the tests, we just want to ensure that all the options work as desired, and that reasonable errors are thrown when invalid options are passed. There's no need to test the actual algorithms themselves, as that should be done in the test for the algorithm

06:37 < RishabhGargGitte> Thanks. I will try writing a few tests and will ask here if I run into any other issue :)

14:11 < AvikantSrivasta4> Hey Mayank, welcome to the community

14:11 < AvikantSrivasta4> You can follow this link

14:11 < AvikantSrivasta4> <https://www.mlpack.org/doc/mlpack-git/doxygen/build.html|https://www.mlpack.org/doc/mlpack-git/doxygen/build.html>

14:11 < AvikantSrivasta4> And try to build mlpack from scratch

14:48 < jonpsy[m]> Hey again! I've noticed in the codebase, we often leave the job of checkin arg validity to Optimize()/Train() instead of to ctor

14:51 < jonpsy[m]> oh and one more, i think ctor delegating is c++11 feature, wouldnt be a cool feature to add? :)

15:06 < RishabhGargGitte> Hii, I was skimming through the Decision tree codebase and I realized that we have only a `Classify` function for trees. Does this mean that they are meant only for classification and not for regression ? Or did I overlook something ?

15:18 < rcurtin[m]> Rishabh Garg (Gitter) : right, our decision tree implementation is classification only. I think there is an issue open about extending that support to regression too

15:22 < RishabhGargGitte> I see. Can you please point me to the issue? Thanks :)

15:27 < zoq[m]> <jonpsy[m] "Hey again! I've noticed in the c"> By doing it inside Optimize()/Train() allows the user to modify parameters after object instantiation. This also means I can test different method settings without copying the object and still have a validation.

15:28 < jonpsy[m]> Interesting..

15:29 < jonpsy[m]> But isn't this possible with ctor validation as well? After instantiation, I can always use setter to change it's value

15:30 < jonpsy[m]> + added bonus of class not created on nonsensical args

15:32 < zoq[m]> Sure you can change values, but in this case you skip the validation step since you moved it into the ctor, right?

15:32 < jonpsy[m]> Hm

15:33 < jonpsy[m]> Because say we're calling Train multiple times. It'll do the checks that many times

15:34 < jonpsy[m]> Ah, that means id have to put conditions in setters

15:34 < jonpsy[m]> Wouldn't that be better tho? The ctor could call the setters as well. There, fixed it!

15:40 < zoq[m]> The issue here is that you would have to change the setter interface from int& size() { return size; } to void size(int s) {}. And lose the convince of changing a setting with obj.size() = 3.

15:42 < jonpsy[m]> I see

15:44 < zoq[m]> If you see a valuable benefit from putting validation into the ctor, please feel free to open an issue for discussion. Nothing is set in stone.

15:45 < jonpsy[m]> Thanks! I was wondering, about the mail I sent on adding algorithm or refactoring. What would be better?

15:50 < zoq[m]> Both are good maybe they can be combined? I guess depends on what algorithm you would select, if it’s a complex one, better to focus on one thing.

15:51 < jonpsy[m]> Glad you think that way ^_^. I want keep the scope lesser to produce a quality gsoc project.

15:53 < jonpsy[m]> Perhaps only 1 algo(SPEA )+ refactoring.

15:54 < zoq[m]> <jonpsy[m] "Perhaps only 1 algo(SPEA )+ refa"> Might be a solution if you go for it make sure to point out what you are planning to do first.

15:55 < jonpsy[m]> Refactoring definitely.

15:56 < jonpsy[m]> Oleksandr Nikolskyy hi oleskandr, check this out. A good way to start can be to build mlpack from source [https://www.mlpack.org/doc/mlpack-git/doxygen/build.html](https://www.mlpack.org/doc/mlpack-git/doxygen/build.html) you can also look at [https://www.mlpack.org/gsoc.html](https://www.mlpack.org/gsoc.html) for more information on GSoC.

15:57 < zoq[m]> > <@_slack_mlpack_U01RJC8AB2M:matrix.org> Hi everyone, I am Oleksandr, just finished my Bachelor of Science in Math & CS at the University of Bonn.

15:57 < zoq[m]> > I am looking forward contributing and learning things!

15:57 < zoq[m]> Welcome, let us know if you have any questions. Btw. Oleksandr that is a neat name don’t think I heard that one before.

15:58 < zoq[m]> <jonpsy[m] "Refactoring definitely. "> Makes sense, yes.

15:59 < jonpsy[m]> I'll have to think long and hard for this. I'll create an issue and tag you? We can discuss there :)

16:00 < zoq[m]> <jonpsy[m] "I'll have to think long and hard"> Sounds good.

16:06 < zoq[m]> > <@_slack_mlpack_U01RJC8AB2M:matrix.org> Thank's! I'll definitely keep in touch.

16:06 < zoq[m]> > It's Ukrainian variant of the Greek name Alexander.

16:06 < zoq[m]> Ahh, I didn’t knew that, cool.