05:56 < mulx10> hello !

05:56 < mulx10> rcutin: can you please review my PR #1734

05:56 < mulx10> thank you

06:06 < rcurtin> mulx10: hi there, I review PRs when I am able to. please be patient

06:06 < rcurtin> I am out of town this weekend

07:22 < sreenik> Does mlpack provide for image augmentation during training/testing ?

09:02 < jenkins-mlpack2> Project docker mlpack nightly build build #244: STILL UNSTABLE in 3 hr 48 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/244/

10:50 < mulx10> rcurtin: Sure, thank you

13:15 < ShikharJ> sreenik: I'm not entirely sure what you mean by that? Could you elaborate?

15:41 < sundar> Hello! I am Sundar, I am a 3rd year Computer Science Student. I am very familiar with the basics of C++ and STL functions in C++14( basic functionalities). I would like to contribute to mlpack in GSOC2019.

15:41 < sundar> I have seen downloaded and built mlpack on my laptpp

15:41 < sundar> *laptop

15:42 < sundar> I have seen an issue, but i am not sure how to proceed with it. Should I ask in GitHub page or here?

15:42 < sundar> Thanks a lot in advance'

16:23 < zoq> sundar: Hello, let's ask on Github, that way we can keep track who is work on what.

16:23 < sundar> yeah okay sure

17:26 < sundar> I have commented in the github page, and have chosen the issue SpikeSlab RBM tests implementation.

17:29 < sundar> I would like to work on the project Essential Deep Learning Modules for GSOC, as I am very interested in Deep Learning.

17:30 < akfluffy> Hey, I have a question regarding the RNN implementation in the library: why do they need cubes as inputs instead of matrices? If each slice is a time slice, and each column is a data point, then why would I need multiple columns in one slice?

18:24 < zoq> akfluffy: Makes it easier to work with multidimensional inputs.

18:27 < akfluffy> Ah. So if I wanted to Predict() 7 elements, the amount of columns in one slice would be 7?

18:27 < akfluffy> One time slice*

18:27 < zoq> right

18:28 < akfluffy> And the labels would be the timestamp correct? If I was predicting over time

18:29 < zoq> if you are going to predict the timestamps right

18:29 < akfluffy> Well, predict a certain feature using timestamps

18:31 < akfluffy> For example, if I wanted to predict a noisy sine, the label would be X right? So X would necessarily be the timestamp

18:32 < zoq> In this case yes.

18:32 < akfluffy> Thank you!

18:35 < zoq> ShikharJ: Do you use the GitHub interface to merge PR's? I got way more mails than commits in the PR.

18:37 < zoq> ShikharJ_: 62 mails

18:44 < zoq> ShikharJ_: I see that one PR did two merges commits against master, so that explains the number.

18:45 < ShikharJ> zoq: Yes, am I doing something wrong?

18:46 < ShikharJ> By GitHub interface I assume you mean the merge button?

18:46 < ShikharJ> I just use the "create an additional merge commit" option.

18:48 < Siddhant> Hello everyone! I am Siddhant , 3rd year Btech CSE student. I want to contribute to mlpack this year in the project idea (mlpack-TensorFlow Translator).

18:48 < Siddhant> I have a basic idea on mlpack

19:05 < jenkins-mlpack2> Project mlpack - git commit test build #114: STILL UNSTABLE in 43 min: http://ci.mlpack.org/job/mlpack%20-%20git%20commit%20test/114/

19:05 < jenkins-mlpack2> noreply: Merge pull request #1728 from dtimson/training-results

19:09 < zoq> ShikharJ: No, everything is okay, it was just the PR.

19:09 < zoq> Siddhant: Hello and welcome.

23:55 < akfluffy> Hey, I have a question: how is it possible that my RNN is predicting negative values when all of the data I trained it on was positive?