naywhayare 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/
< jenkins-mlpack>
Starting build #1960 for job mlpack - svn checkin test (previous build: SUCCESS)
< jenkins-mlpack>
* Ryan Curtin: For clarity, use separate split and binLabels objects instead of storing the label in the split matrix. No casting is necessary anymore.
< jenkins-mlpack>
* Ryan Curtin: Use const double where possible instead of having a variable used throughout the
< jenkins-mlpack>
function. This is only for clarity and consistency and is not likely to make
< jenkins-mlpack>
any runtime difference.
< jenkins-mlpack>
Starting build #1961 for job mlpack - svn checkin test (previous build: SUCCESS)
< jenkins-mlpack>
Ryan Curtin: Remove unused variables.
Anand_ has joined #mlpack
< Anand_>
Marcus : I will try implementing logistic regression for weka/scikit. The task at hand is to get the probs for mlpack so that mlpack logistic regression is completed. i need your help there. After that I will start implementing LR for other libraries.
< Anand_>
After that I will take HMMs
Anand_ has quit [Ping timeout: 246 seconds]
Anand_ has joined #mlpack
< jenkins-mlpack>
Starting build #1962 for job mlpack - svn checkin test (previous build: SUCCESS)
< jenkins-mlpack>
Ryan Curtin: Oops, missed one unused variable.
sumedh_ has joined #mlpack
< sumedh_>
naywhayare: hey ryan, U awake? Done with the abstraction. Just want to confirm with you before commit.
udit_s has joined #mlpack
< udit_s>
naywhayare: are you there ?
< udit_s>
marcus_zoq: are you free now ?
udit_s has quit [Quit: Leaving]
andrewmw94 has joined #mlpack
sumedh__ has joined #mlpack
sumedh_ has quit [Ping timeout: 255 seconds]
govg has joined #mlpack
govg has quit [Changing host]
govg has joined #mlpack
govg has quit [Quit: leaving]
govg has joined #mlpack
Anand has joined #mlpack
govg has quit [Ping timeout: 255 seconds]
govg has joined #mlpack
govg has quit [Changing host]
govg has joined #mlpack
< sumedh__>
naywhayare: Hey ryan I figured out the the problem with cmake and header file.. If header file is modified or new header file is added the header file does not get copied... Its because of the use of file(GLOB.. its given online this use is not recommended
< naywhayare>
oh! cool. do you have a fix?
< naywhayare>
if so, please commit it and resolve ticket 322
< sumedh__>
yeah... I am right now doing that... like you add all the sources to MLPACK_SRCS ...
< sumedh__>
add all the includes to another variable
< sumedh__>
MLPACK_INCLUDES
< sumedh__>
do you know how to get relative path from absolute path??
< sumedh__>
any string subtraction function??
< naywhayare>
that is not a good solution, that is a huge amount of code; there must be an easier way
< sumedh__>
yeah I know... bt the problem is file(GLOB.. does not track files... when a new file is added it won't copy that file unless you touch that CMakeList file with new timestamp
Anand has quit [Ping timeout: 246 seconds]
< naywhayare>
all of the header files are already collected with the MLPACK_SRCS variable, so you can parse that list for any hpp files
< naywhayare>
also, bug 322 is a different problem -- the header files aren't being copied when they are modified
< naywhayare>
that's different than the problem the cmake documentation is detailing, which is when a header file is added or removed but no changes are made to the CMake configuration
< naywhayare>
basically, the issue is that every single time the mlpack target is being built, the header files should be moved
< naywhayare>
*the header files should be copied to build/include/
< sumedh__>
how to parse that list for .hpp files??
< naywhayare>
I'm not sure; read the CMake documentation. probably some regex magic is what is necessary
andrewmw94 has quit [Remote host closed the connection]
andrewmw94 has joined #mlpack
udit has joined #mlpack
< marcus_zoq>
udit: Hello, do you need some help?
< udit>
yeah, I was having some trouble while I was going through adaboost literature.
< udit>
marcus_zoq: I was wondering how training a weighted classifier would work.
< udit>
marcus_zoq: I've basically understood how to implement adaboost should be implementd, except for the training the weighted classification.
< udit>
as in how they would translate for the decision stump and perceptron.
udit is now known as udit_s
< udit_s>
marcus_zoq: because that might involve rewriting the API for these two.
udit_s has quit [Quit: leaving]
< marcus_zoq>
udit_s: The aim of the weak learner (perceptron/decision stump/naive bayes) is to find a weak hypothesis with low weighted error given a training example. After each classifier is trained, the classifier's weight is calculated based on its accuracy. We give a classifier with a high accuracy more weight.
< sumedh__>
naywhayare: I tried some solutions... I have a working solution but one has to run "cmake" before running make to reflect the modifications...
< naywhayare>
sumedh__: that's not a solution that will work, then. :( we can't expect users to type 'cmake' every time before they type 'make'
< naywhayare>
unless I've misunderstood what you meant
< sumedh__>
yeah... I haven't given up yet :)
udit_s has joined #mlpack
udit_s has left #mlpack []
< jenkins-mlpack>
Starting build #1963 for job mlpack - svn checkin test (previous build: SUCCESS)
< sumedh__>
naywhayare: to get all the includes in INCLUDES_FILES can be done with simple regular expression matching to MLPACK_SRCS...
< sumedh__>
code for that...
< naywhayare>
right; now we should make the 'mlpack' target dependent on this target
< naywhayare>
I think it might be possible to do this without the tar... just ${CMAKE_COMMAND} -E copy, maybe
< sumedh__>
naywhayare: is the dependency required?? cause the target is running everytime anyway...
< naywhayare>
but if you type 'make mlpack', that target probably won't be built
< sumedh__>
yeah.. right...
< sumedh__>
I looked for such copy command... I didn't find any...
< sumedh__>
naywhayare: in the meantime... that abstraction is ready... building fine... added 3 file... simple_residue_termination.hpp, simple_tolerance_termination.hpp and validation_rmse_termination.hpp...
< sumedh__>
how to run nmf_test now??
< naywhayare>
there is definitely a ${CMAKE_COMMAND} -E copy
< naywhayare>
you should use that instead
< sumedh__>
yes but it only copies one file...
< sumedh__>
how to copy multiple files??
< naywhayare>
use a for loop
< sumedh__>
then multiple targets have to be added...
< sumedh__>
a target for each file...
< naywhayare>
use add_custom_command then
< sumedh__>
then how to add all those command to target??
< sumedh__>
sorry I am not getting you :(
< naywhayare>
read the documentation for add_custom_command. it allows you to add a command to a target
< naywhayare>
you could add individual copy commands for each file to the mlpack target
< naywhayare>
sumedh__: yes, what you wrote is what I meant. that is currently how the code is. however, those commands are not always executed when the mlpack target is built
< naywhayare>
which is why the ticket is open in the first place
< naywhayare>
maybe if you create a different custom target, which is always out of date, then the custom commands will always be run
< sumedh__>
yes... I read that online but also that does not work...
< sumedh__>
naywhayare: before that... how to run a specific test.. like nmf_test alone??
< naywhayare>
the command 'add_custom_target()' creates a target that " is ALWAYS CONSIDERED OUT OF DATE even if the commands try to create a file with the name of the target."
< naywhayare>
you should use that function to find a way to make the commands run every time that the mlpack target is built
< sumedh__>
so lets try this way... right now all custom commands are linked with mlpack ... i will link them to new custom target instead... and add that target as dependency... will that work??
< naywhayare>
if the cmake documentation is not lying, then it seems like it might
< sumedh__>
naywhayare: okay all tests are passing with termination policy... making the commit
< jenkins-mlpack>
Starting build #1964 for job mlpack - svn checkin test (previous build: SUCCESS)