verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
topology has joined #mlpack
< topology> rcurtin: i am sorry i couldn't complete what i was saying yesterday.
< topology> in absolute valued approximate KDE, we require that the difference between our approximation and the true density estimate for each point be less than some parameter 'e'.
< topology> from this, i can understand why we should prune when the bound(for the maximum difference between 2 kernel values for a given combination of nodes) B >= e/|Sr|
< topology> where Sr is the reference set.
< topology> this is what is used in the ICML 2013 paper as well.
< topology> but the KDE score() function in your thesis prunes when the same bound is less than the parameter 'e'.
< topology> i am not able to understand how we reached that condition?
thyrix has joined #mlpack
topology has quit [Ping timeout: 260 seconds]
topology has joined #mlpack
delfo has joined #mlpack
huyssenz has quit [Quit: Connection closed for inactivity]
delfo_ has joined #mlpack
thyrix has quit [Remote host closed the connection]
delfo has quit []
mikeling has joined #mlpack
delfo_ has quit [Quit: WeeChat 1.7]
brn_ has joined #mlpack
vivekp has quit [Ping timeout: 256 seconds]
topology has quit [Ping timeout: 260 seconds]
brn_ has quit [Quit: WeeChat 1.7]
brn_ has joined #mlpack
brn_ has quit [Quit: WeeChat 1.7]
brn_ has joined #mlpack
shihao has quit [Quit: Page closed]
brn_ has quit [Quit: WeeChat 1.7]
ravi__ has joined #mlpack
< ravi__> hello
< ravi__> Am i talking to someone from mlpack
< ravi__> Sir
< ravi__> I tired mailing you regarding Gsoc
< ravi__> but it said only subscribers can mail
< ravi__> ??? please help
ravi__ has quit [Client Quit]
vivekp has joined #mlpack
topology has joined #mlpack
dipta has joined #mlpack
dipta has quit [Client Quit]
flyingpot has joined #mlpack
thyrix has joined #mlpack
flyingpot has quit [Remote host closed the connection]
thyrix has quit [Ping timeout: 260 seconds]
ravi__ has joined #mlpack
< ravi__> ?HELP
< ravi__> ?/
ravi__ has quit [Quit: Page closed]
zoq_ has joined #mlpack
zoq has quit [Read error: Connection reset by peer]
thyrix has joined #mlpack
ravi__ has joined #mlpack
< ravi__> Hey! people , seniors , sirs and Coders .... This is Ravi sadhwani ... want to know about GSOC and all ... Can anyone please help me out
ravi__ has quit [Ping timeout: 260 seconds]
vinayakvivek has joined #mlpack
dineshraj01 has joined #mlpack
dineshraj01 has quit [Remote host closed the connection]
Narayan_Srinivas has joined #mlpack
Narayan_Srinivas has quit [Quit: Page closed]
damien has quit [Ping timeout: 256 seconds]
sicko has joined #mlpack
aditya_ has joined #mlpack
thyrix has quit [Ping timeout: 260 seconds]
thyrix has joined #mlpack
vinayakvivek has quit [Quit: Connection closed for inactivity]
govg has quit [Ping timeout: 258 seconds]
govg has joined #mlpack
brn_ has joined #mlpack
zoq_ is now known as zoq
brn_ has quit [Quit: WeeChat 1.7]
dashwood has joined #mlpack
< dashwood> heyy guys i am facing an error while running linear regression code
< dashwood> fatal error: 'mlpack/core.hpp' file not found
< dashwood> which directory should i place the mlpack directory
< dashwood> i am using Mac OS
< zoq> dashwood: You have to build mlpack, it's not header only: http://www.mlpack.org/docs/mlpack-2.1.1/doxygen.php?doc=build.html#build
< zoq> dashwood: Since you are on mac you can probably use homebrew to install all dependencies? There is also an mlpack package in homebrew science, if you just like to use mlpack e.g. linear regression.
< dashwood> i already installed all dependencies
< dashwood> armadillo
< dashwood> and everything thats mentioned
< dashwood> and it has installed it into /usr/include/mlpack
< dashwood> but for some reason my g++ compiler is showing error
< zoq> So I guess what you like to do is to build against mlpack and use the linear regression code in your own code?
< dashwood> yes i want to perform linear regression
< zoq> hm, just asking because you could use the executable that should be located in '/usr/local/bin/mlpack_linear_regression -h' or if you are in 'build bin/mlpack_linear_regression -h'
< zoq> anyway, how does your g++ command look like?
< dashwood> i wrote the code in a test.cpp file...and i am using g++ test.cpp
< zoq> So, yeah you have to build against mlpack something like: g++ test.cpp -std=c++11 -I/path/to/mlpack/build/include/ -L/path/to/mlpack/build/ -lmlpack -larmadillo
thyrix has quit [Quit: Page closed]
< dashwood> fatal error: 'mlpack/methods/linear_regression/linear_regression.hpp' file not found i still get this error
< dashwood> i used this g++ command
< dashwood> g++ test.cpp -std=c++11 -I/usr/local/include/mlpack/build/include/ -L/usr/local/include/mlpack/build/ -lmlpack -larmadillo
< zoq> Who does your test.cpp look like? Maybe you can post it on pastebin?
vinayakvivek has joined #mlpack
< dashwood> one minute
< dashwood> thats the code
< zoq> The link doesn't work for me.
< dashwood> try this
< zoq> okay, that looks good, so can you check if you can find 'linear_regression.hpp' in /usr/local/include/mlpack/build/include/ and 'libmlpack' in /usr/local/include/mlpack/build/?
< zoq> g++ test.cpp -std=c++11 `pkg-config --libs mlpack` might also work
topology has quit [Ping timeout: 260 seconds]
dashwood_ has joined #mlpack
< dashwood_> i got it running there but
< dashwood_> atal error: 'boost/math/special_functions/gamma.hpp' file not found
topology has joined #mlpack
< zoq> What command did you used to get the new error?
< dashwood_> the same one
< dashwood_> turns out boost is installed in some other directory if i use brew install
< dashwood_> boost goes into /usr/local/Cellar/boost/1.63.0/
< dashwood_> should i move it?
< dashwood_> could you give me the command you gave before?
< rcurtin> topology: sorry for the slow answer, I think there is an error in the thesis, so I would go with the ICML version
< rcurtin> to me, it seems like the approximation the thesis version calculates is | f(p_q) - f^*(p_q) | < \epsilon * |S_r|
< rcurtin> not | f(p_q) - f^*(p_q) | < \epsilon
< zoq> dashwood_: Don't move the folder: 'g++ test.cpp -std=c++11 -I/Users/marcus/src/mlpack/build/include -L/Users/marcus/src/mlpack/build/lib -lmlpack -larmadillo' is the one I used to build your code.
< dashwood_> 'boost/math/special_functions/gamma.hpp' file not found
< dashwood_> this is the error i get
< rcurtin> maybe add -I/usr/local/Cellar/boost/1.63.0/ ? but I would imagine that would already be on the include search path
< zoq> rcurtin: About to answer another mail: "What resources can you advice to brush up *advanced/modern* C++? (e.g., templates, && in functions)"
< zoq> I would go with "Modern C++ Design, Generic Programming and Design Patterns Applied" by Andrei Alexandrescu, it's still my favorite. Do you have anything to add?
< rcurtin> I'm trying to think of the name...
< rcurtin> Scott someone wrote a few books on the STL
< zoq> Scott Meyers?
< rcurtin> I want to say Scott Manley but that's the wrong guy, he just publishes videos about Kerbal Space Program on youtube
< rcurtin> yes!
< rcurtin> Scott Meyers
< zoq> the effective series I guess
< rcurtin> Effective C++ and Effective STL are good books
< rcurtin> yeah
< zoq> yeah, I agree "Effective C++" is also nice
< zoq> thanks!
thyrix has joined #mlpack
< rcurtin> I guess, maybe we should add those books as suggestions to gsoc.html?
< zoq> Might be helpful to dive into some cool aspects.
deepanshu_ has joined #mlpack
< rcurtin> sure, I will update the page
witness_ has joined #mlpack
dashwood_ has quit [Ping timeout: 260 seconds]
dashwood has quit [Ping timeout: 260 seconds]
< mikeling> hi rcurtin , I got "for declaration does not refer into a class, class template or class template partial"if I define class like https://pastebin.mozilla.org/8981310
< mikeling> I still fell confuse after I google it
travis-ci has joined #mlpack
< travis-ci> mlpack/mlpack#1985 (master - 5215c3d : Marcus Edel): The build is still failing.
travis-ci has left #mlpack []
< rcurtin> mikeling: is there a forward definition of that method in decision_tree.hpp?
< rcurtin> make sure it matches what you've got implemented in decision_tree_impl.hpp
thyrix has quit [Quit: Page closed]
benchmark has joined #mlpack
benchmark has quit [Client Quit]
< topology> rcurtin: what do you think would be an appropriate deliverable for a proposal?
< topology> is implementing the simple KDE algorithm first during the summer enough? we can then build on it using Bill's paper and other approaches.
< topology> or is too little work for a gsoc project?
Nax has joined #mlpack
Nax has quit [Ping timeout: 260 seconds]
mikeling has quit [Quit: Connection closed for inactivity]
< rcurtin> topology: I think that the simple KDE algorithm would not be enough for a full summer; because mlpack already has this nice dual tree algorithms infrastructure, honestly I think it could be implemented in a week or two
< topology> rcurtin: i thought as much
< topology> i think i should begin reading Bill's paper for ideas now. what do you think?
< rcurtin> that could be a good idea, yeah, but I am not sure that he presents the algorithms in a tree-independent way, so you may have to do some "translation"
< rcurtin> another idea might be to implement the simple KDE as a test to make sure you get the dual-tree algorithm abstractions that are in mlpack
< rcurtin> alex gray's paper from 2003 SDM might be good to look at too, he lays out a complex algorithm to be used with kd-trees
aditya_ has quit [Ping timeout: 240 seconds]
topology_ has joined #mlpack
< topology_> that sounds like a good idea. i do theoretically understand the dual-tree abstraction as explained in your thesis.
< topology_> But the actual code can sometimes be a little obtuse. so i think you are right, it would be a good idea to implement the simple KDE first.
< topology_> i will begin working on it as soon as i finish going through Bill March's and Alex Gray's papers. (i want to make sure i have enough to put up a strong proposal first)
< topology_> does it sound good?
topology has quit [Ping timeout: 260 seconds]
topology_ is now known as topology
< rcurtin> yes, of course, that seems reasonable to me
< rcurtin> remember that the overarching goal in the end would be to provide a KDE program users can use that is both fast and accurate
< rcurtin> there are many papers out there about doing KDE fast (many of them using single-tree or dual-tree algorithms), so you should be familiar with the literature and which ones you'll aim to implement
< rcurtin> "ones" or "one" depending, some of them are quite complex
< rcurtin> but definitely the proposal should at least have the basic dual-tree (and single-tree) KDE as a starting point
< topology> i understand that. i will read the papers and get back to you for clarifications and further discussions :)
< rcurtin> sure, sounds good
< sicko> rcurtin, Hello, just wanted to say thanks. :)
outofnames has joined #mlpack
< rcurtin> sicko: sure, I am not sure what I did, but you are welcome :)
sicko has left #mlpack []
govg has quit [Ping timeout: 240 seconds]
topology has quit [Ping timeout: 260 seconds]
outofnames has quit [Quit: Page closed]
govg has joined #mlpack
kris1 has joined #mlpack
< kris1> arunreddy: Good work on momentum+sgd pr. Though i just gave it glance .....:)
< arunreddy> krsi1: Thanks :)
googleson78 has joined #mlpack
< rcurtin> sicko: sure, I am not sure what I did, but you are welcome :)
< rcurtin> oops, wrong terminal to hit "up-enter" in
< kris1> zoq: layer->Parameters() return 21*1 matrix so basically a row vector.
< kris1> shoudn't it return matrix that should be equal to input_size*output_size
< kris1> okay sorry i think i got it
< kris1> is the input layer model.Model()[0] ?
aa_ has joined #mlpack
aa_ has quit [Ping timeout: 260 seconds]
vinayakvivek has quit [Quit: Connection closed for inactivity]
witness_ has quit [Quit: Connection closed for inactivity]
deepanshu_ has quit [Quit: Connection closed for inactivity]
< zoq> kris1: model.Model()[0] is the input/first layer, right.