06:05 < wenhao> Hi, does anyone know of an armadillo function like transform() or for_each() which can modify only nonzero data in a sparse matrix?

06:08 < wenhao> transform() and for_each() are supported for sparse matrix since arma 8.5, but mlpack should support armadillo versions since 6.5

06:24 < sourabhvarshney1> zoq: Sorry for the delay again. A tragedy happened. So, cannot open the pr. Will open it today for sure.

06:25 < sourabhvarshney1> zoq: Can you provide me with boost::apply visitor method explanation. I am not getting clear with the usage.

06:33 < zoq> wenhao: We could backport the necessary features, see https://github.com/mlpack/mlpack/tree/master/src/mlpack/core/arma_extend

06:33 < zoq> wenhao: But I think Ryan can point us in the right direction since he basically wrote a bunch of the armadillo sparse matrix support.

06:37 < zoq> sourabhvarshney1: No worries, helpfully everything is okay.

06:37 < zoq> sourabhvarshney1: http://klemens-morgenstern.github.io/visitor/index.html#boost_visitor.introduction.abstract.visitor_motivation should be helpful, the key word here is boost visitor pattern.

07:04 < sourabhvarshney1> zoq: So, I think I am clear now. A visitor is basically like a overloaded functor which applies operation according to type of argument. Am I right?

07:28 < sourabhvarshney1> zoq: I need to create two visitors for the forward as well as the backward network for example there will be two deltas. So, I think I would need to create more visitors as all the visitors directly call the setter which would be different in our case as there will be two different setters and we need to call them accordingly.

08:29 < wenhao> zoq: Thanks! sp_mat::iterator works but may not be as safe and efficient as transform() and for_each(). Let's wait for Ryan then

09:50 < jenkins-mlpack> Project docker mlpack nightly build build #323: STILL UNSTABLE in 2 hr 36 min: http://masterblaster.mlpack.org/job/docker%20mlpack%20nightly%20build/323/

12:08 < ShikharJ> zoq: Out of curiosity, do you have any update regarding the stickers that you mailed me back in February? I think I should have received them by now (international mailing take about 4-6 weeks on standard).

14:35 < zoq> ShikharJ: Ohh, I think you should have received them by now. I know that Haritha already received the stickers, not sure I can get any information from the post office, but let's see, if not I can send you another package.

16:44 < manish7294> rcurtin: I made a rough implementation of lmnn, but results are quite depressing. On iris dataset, at k = 6, out of 378 points --- original dataset using knn is able to successfully classifying 357 points whereas after learning metric--- the number changed to 348, Most probably I am making some kind of mistake :(

16:45 < manish7294> Thought, I should update you with the current progress.