ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
travis-ci has joined #mlpack
< travis-ci> mlpack/examples#762 (master - b10d1d6 : Ryan Birmingham): The build has errored.
travis-ci has left #mlpack []
< AakashkaushikGit> Hey @zoq, I have created a PR in you branch ann-vtable after porting the `rnn` class but i am think it is not actually compiling it even after i un commented it from the cmake file, can you please take a look ?
< AakashkaushikGit> (edited) ... i am think it is not actually compiling it even after i un commented it ... => ... i think it is not actually compiling it even after i un-commented it ...
< AakashkaushikGit> Also if someone gets time can they give their opinions on this [proposal](http://knife.lugatgt.org/pipermail/mlpack/2021-March/004452.html) i sent to the mailing list, I know it's been really busy because of GSOC, i will be glad if someone can take a look on this.
ImQ009 has joined #mlpack
Agzam has joined #mlpack
Agzam has quit [Client Quit]
Hao has joined #mlpack
Hao has quit [Quit: Connection closed]
lazaros65 has joined #mlpack
lazaros65 has quit [Client Quit]
amber51 has joined #mlpack
amber51 has quit [Client Quit]
< shrit[m]> would it throw an error if you use Test<false> directly?
< RishabhGargGitte> No, it wont throw an error.
< MohomedShalik[m]> template instantiation happens in compile time, usebool seems to be decided at run time
< RishabhGargGitte> Yep, you are correct @Mohomed , so, i guess I should write an if-else statement for useBool = true and useBool = false. Thanks :)
< zoq> AakashkaushikGit: Thanks, I saw the commit, I'll take a look.
< AakashkaushikGit> @zoq thanks
< jonpsy[m]> > But this is throwing me an error. Can someone tell me another way to do this? All I want is to call a function template such that the template parameter is decided by a variable.
< jonpsy[m]> Rishabh Garg (Gitter): templates work in compile time, so all your variables should be known during compile stage
< jonpsy[m]> you can try [constexpr](https://en.cppreference.com/w/cpp/language/constexpr) perhaps?
< jonpsy[m]> zoq: Hi, do you think MOEA/D-DE is GTM? or will it take a month or so. Wanted to know so that I can add it in GSoC :) (and drop NSGA-III)
< zoq> jonpsy[m]: It's on my list for today, I'll let you know.
< jonpsy[m]> sure, thanks
< RishabhGargGitte> Thanks @jonpsy , actually I changed that template variable to a function argument. Templating it was causing these unnecessary problems so I decided to switch it. :)
< jonpsy[m]> Rishabh Garg (Gitter): I see, but i still recommend reading constexpr, cool stuff really!
< RishabhGargGitte> Sure, will look at it too :+1:
ib07 has joined #mlpack
< kaushal[m]> <brongulusGitter[ "Install the `tqdm` module for py"> did already still showing the issue
< zoq> kaushal[m]: Make sure you installed it for the correct python version (python2 or python3), you can check to import tqdm in python manually.
< jonpsy[m]> is there to get max along each axis, for a matrix, returning a max vector?
< jonpsy[m]> ```arma::mat a(2,3); a.max(axis=1)```, something like this..
< rcurtin[m]> yeah, the `max()` function for Armadillo does allow you to specify a dimension on which to compute the maximum: http://arma.sourceforge.net/docs.html#min_and_max
< jonpsy[m]> cool!, on the side, how can someone make PR to armadillo itself?
< jonpsy[m]> Ah thanks! In future if I want to add an awesome feature, I could put it there ^_^
< rcurtin[m]> yep :)
< kaushal[m]> everything is installed
< zoq> kaushal[m]: Can you check if you can download http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz ?
< zoq> It's about 9.5MB.
< kaushal[m]> it shows service service unavailabe
< kaushal[m]> it shows service unavailable
< zoq> kaushal[m]: Are you behind a firewall?
< kaushal[m]> no I don't think so
< zoq> kaushal[m]: Ohh I now get the same message.
< zoq> I guess let's wait and try again in 30 minutes or so.
< zoq> Nothing we can do really.
< kaushal[m]> zoq: yes well what is it
< zoq> "The server is temporarily unable to service your request due to maintenance downtime or capacity problems. Please try again later.
< kaushal[m]> <zoq ""The server is temporarily unabl"> zoq:no I know that , but i was asking what is this link
< zoq> It's the dataset you are downloading.
< kaushal[m]> zoq: anyways i was trying to run this examples directory and contribute to it in the GSOC'21
< kaushal[m]> zoq:
< zoq> What I did I just followed the error message you posted and checked if I can manually download the file.
< kaushal[m]> zoq: I was wondering if we could use audio dataset for clustering algorithms
< zoq> I like the idea, I wanted to use kmeans and knn to cluser bird sounds, but never got to the point to write an example for that.
< kaushal[m]> zoq: you think it would be a good idea for Example zoo gsoc'21 idea ?
< zoq> Doing something with audio, sure.
< zoq> The tricky part is to extract the features.
< kaushal[m]> I just read a paper on audio noise clustering and it seems quite interesting but tricky
< zoq> rcurtin[m]: Did you use mlpack to classify broiler chicken?
< rcurtin[m]> no, it was MATLAB 😭
< zoq> kaushal[m]: Yeah, there are some cool audio datasets.
< zoq> rcurtin[m]: Okay too bad.
< kaushal[m]> zoq: I don't think there are plotting tools in mlpack so can i use octave for visualization
< zoq> kaushal[m]: There are two things I would take a closer look at first, how can you load the audio files, and what methods do you need to extract features.
< kaushal[m]> <zoq "kaushal: There are two things I "> yeah well there are two ways of loading audio datasets
< kaushal[m]> 1: spectrogram
< zoq> kaushal[m]: I would use xplot - https://github.com/QuantStack/xplot
< kaushal[m]> 2: 3d-table
< zoq> kaushal[m]: And do it inside a notebook
< kaushal[m]> <zoq "kaushal: And do it inside a note"> this QuantStack/xplot right?
< zoq> kaushal[m]: the two methods are not for loading a wav dataset
< kaushal[m]> <zoq "kaushal: the two methods are not"> its what i saw in the paper spectrogram(frequency bins over time)
< zoq> Yeah, but I was talking about how you can load the wav data, you can't use mlpack::data::Load(...)
< kaushal[m]> <zoq "Yeah, but I was talking about ho"> yeah the spectrogram would make matrix , but i am also not so sure on this
< kaushal[m]> maybe we can find a way to load the audio data into spectrogram and then convert it into matrix of arma::mat type
< kaushal[m]> I will look onto this
< zoq> kaushal[m]: Yeah, good idea to get familiar with the methods first.
< kaushal[m]> I think only the loading of data is the problem
< kaushal[m]> the implementation gets a bit easier after this
ImQ009 has quit [Quit: Leaving]
gotadachi has quit [Read error: Connection reset by peer]
< rcurtin[m]> zoq: do you know any way to debug kernel failures on binder? when I try to run the pima diabetes notebook, the kernel dies on `auto im = xw::image_from_file("output.gif").finalize();`, but I don't know any way to get any more output than that...
gotadachi has joined #mlpack
< zoq> rcurtin[m]: On binder you can't really, xw can be picky sometimes, often if you restart the whole notebook again it runs just fine.
< rcurtin[m]> ah, ok, let me try that...
< zoq> I can take a look as well, I got the mlpack binder instance running again.
< zoq> rcurtin[m]: Do you have a link to the dataset at hand?
< zoq> rcurtin[m]: nevermind, I can reproduce the issue, I'll fix it later today.
< zoq> rcurtin[m]: I can create the screenshot for you.
< zoq> rcurtin[m]: Or if you like you can remove the xw::image_from_file like for now and add a new code line and use markdown.
< zoq> e.g. ![image](pca.png)
< zoq> you just have to change the interpreter for the line from code to markdown.