12:57 < Soonmok> Hi ! I try to import mlpack library to study it. but i found it difficult to link it using Cmake.

12:58 < Soonmok> i already succeed to build mlpack using CMake, Since my dev environment is mac, i am using Clion.

13:00 < Soonmok> but I can't find any tutorial for mac user.

13:07 < Soonmok> I typed 'find_package(mlpack)' in CMakeList.txt in my project directory

13:07 < Soonmok> but it causes this error ' CMake Warning at CMakeLists.txt:12 (find_package): By not providing "Findmlpack.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "mlpack", but CMake did not find one. Could not find a package configuration file provided by "mlpack" with any of the following names:'

13:11 < Soonmok> When i type 'mlpack_knn --help' command in terminal, the help docs is displayed

13:26 < Soonmok> whooooaaaa i succeed to add lib!!

14:15 < rajiv_> Hello developers, is there an interesting deep learning problem to work on which you might like to suggest?

14:15 < rajiv_> I am extremely interested to work on it

15:59 < amet_> can anyone tell me how to do tests once made your changes in src/mlpack/methods/<somefile> ?

15:59 < amet_> ^particularly for that method where you made change

16:18 < niteya> rcurtin: I made the changes , can you verify them ? (PR #1632)

16:18 < rcurtin> niteya: sure, I will review the PR shortly

16:22 < niteya> amet_: All the tests are built as a executable in the bin - mlpack_test , to execute a single test , say KNNTest , mlpack_test -t KNNTest

16:23 < amet_> ok thanx

16:26 < rcurtin> ShikharJ: zoq: I think I'm going to sit down tonight and come up with a new labeling scheme for issues

16:26 < rcurtin> and maybe try to set up a github bot to auto-label issues, close stale issues, and to offer to mail stickers to new contributors :)

16:27 < rcurtin> I've found in the past handful of days that it's just overwhelming to keep up with all of the issues that come in every day

16:27 < rcurtin> and it's very difficult for me to actually make any forward progress in the development of the library

16:27 < rcurtin> so, I think, at least for me, I'd like to experiment with new strategies of managing what's going on

16:29 < rcurtin> if either of you are interested in helping me figure out some ideas, I can try and send an email with a draft of the ideas tonight

16:29 < rcurtin> otherwise I can just do it and we can see what we think :)

16:47 < niteya> rcurin: I had added the serialisation and read dataset features in my last commit .

17:04 < saurabh97> Hi everyone , I was looking around Gradient boosting and didnt find any support for the same in mlpack , is boosting available in mlpack?

17:06 < rcurtin> saurabh97: did you look at adaboost? that is a boosting algorithm, but it is not, e.g., gradient-boosted decision trees

17:10 < saurabh97> yeah sorry I meant gradient boosted trees in particular

17:13 < rcurtin> right, those are not implemented, but would be nice to add for sure

17:21 < saurabh97> okay thanks

18:27 < zoq> rcurtin: Same, happy to help and to reduce the backlog :)

18:29 < zoq> Soonmok: Nice to hear you figured it out, always wanted to take a closer look into Clion.

18:34 < rajiv_> rcnn and capsule nets?

18:37 < zoq> rajiv_: That is an option :)

18:47 < rajiv_> Region of interest pooling layers are gaining a lot of attention

18:54 < rajiv_> Auto encoders

19:19 < rcurtin> zoq: sounds good, I'll send a ping tonight

19:21 < zoq> rcurtin: Sounds good.

19:41 < ShikharJ> rcurtin: And maybe an additional templated comment when someone opens a PR. For example, something along the lines of this (https://github.com/sympy/sympy/pull/15792), and mentioning whom to ping for certain issues. Like it could be "For changes to the ANN code, please ping @ShikharJ" or something.

19:50 < rcurtin> ShikharJ: that wouldn't be a bad idea; definitely a bot to help management of issues is something that's needed

19:58 < zoq> I'm not sure there is any need to make a specific recommendations (github already does this) and I don't think we should create the impression that only a subset of the community is able to provide feedback.

20:01 < rcurtin> yeah, the 'reviewer suggestions' seems to be pretty good already

20:01 < rcurtin> I think it just uses 'git blame'

20:08 < zoq> Not sure, did someone from the community receive a review request? Personally, if I have the chance to provide feedback I do that, nobody has to send me a request, I get a notification anyway.

20:14 < rcurtin> same here, although I am finding I'm basically at the limit of what I am able to do these days

20:14 < rcurtin> and I suspect that as time goes on, the rate of PRs and issues opening will increase :)

20:24 < zoq> Agreed, we can definitely do something that helps with the amount of work.

20:52 < ShikharJ> zoq: Hmm, of course we don't want to do that, but it is borderline impossible for me to take a look at all PRs. I had unwatched the repository previously, but most of the people do not end up pinging me, and I can't really open up and start looking at PRs one by one to check if there is any that changes the ANN code in a certain way or not. So I put the repository to watch again, which was fine until now, but the number of

20:52 < ShikharJ> mails is getting out of hand :/

20:58 < rcurtin> ShikharJ: agreed. I think, no matter what the workflow ends up looking like for each of us, there will have to be a step where everyone looks through the issues/PRs to find what they are interested in

20:58 < rcurtin> but I think with more automated labeling, this can get a little bit easier

21:27 < zoq> Agreed meaningful labels are a good step to help everyone to filter topics of interest. And it should be straightforward to send notifications to people who subscribed to labels if that is something someone likes to get.

23:59 < ShikharJ> zoq: lozhnikov : I was interested in mentoring for Essential Deep Learning Modules project, and possibly co-mentoring alongside you guys. Would you be interested in the same?