IRC logs of #shogun for Wednesday, 2011-12-07

--- Log opened Wed Dec 07 00:00:16 2011
-!- blackburn [~blackburn@188.168.128.146] has quit [Quit: Leaving.]00:07
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun03:26
-!- k_matsuda [8370b643@gateway/web/freenode/ip.131.112.182.67] has joined #shogun04:14
k_matsudahi, i considering modshogun ruby bindings to perform MKL.04:15
k_matsudabut i have a problem about ruby binding04:16
k_matsudai initialiized Modshogun::CombinedKernel.new()04:16
k_matsudaand combined_kernel.append_kernel(individual_kernel)04:17
k_matsudato perform lerning, Modshogun::MKLClassification.new(); mklsvm.set_kernel(combined_kernel)04:17
k_matsudaand set labels by mklsvm.set_labels(Modshogun::Labels.new([1,-1]))04:18
k_matsudai expect MKLClassification#get_kernel will return CombinedKernel object04:20
-!- k_matsuda [8370b643@gateway/web/freenode/ip.131.112.182.67] has quit [Ping timeout: 265 seconds]04:24
-!- k_matsuda [~k_matsuda@peter.pi.titech.ac.jp] has joined #shogun04:28
k_matsudasorry, my network is bad...04:29
k_matsudai initialiized Modshogun::CombinedKernel.new() and combined_kernel.append_kernel(individual_kernel)04:30
k_matsudato perform lerning, Modshogun::MKLClassification.new(); mklsvm.set_kernel(combined_kernel)04:30
k_matsudaand set labels by mklsvm.set_labels(Modshogun::Labels.new([1,-1]))04:30
k_matsudai expect MKLClassification#get_kernel will return CombinedKernel object but it return Kernel object.04:30
k_matsudaso, i cannot get weight of individual kernels by CombinedKernel#get_subkernel_weights method.04:32
k_matsudai try like http://permalink.gmane.org/gmane.comp.ai.machine-learning.shogun/68704:33
k_matsudabut, kernel.get_first_kernel() is not work.04:34
k_matsudabecause get_kernel method return Kernel object (not CombinedKernel)04:34
k_matsudaSo, what to do get weight of individual kernels?04:35
k_matsudaI use shogun git-head (installed yesterday) and use ruby-1.9.204:36
-!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 244 seconds]06:04
sonne|workk_matsuda: I understand the problem08:54
sonne|workyou currently can only work around this by saving the combined kernel object08:55
sonne|workproblem is simply that swig has no notion of casting objects08:55
sonne|workk_matsuda: actually you can use combined_kernel.KernelToCombinedKernel(generic_kernel_obj)09:00
sonne|workwe added that very recently...09:01
k_matsudathanks for your advice!09:18
k_matsudai'did like this:09:19
-k_matsuda:#shogun- # cast learned Kernel obj to CombinedKernel09:20
-k_matsuda:#shogun- casted_kernel = combined_kernel.KernelToCombinedKernel(mklsvm.get_kernel)09:20
-k_matsuda:#shogun- w = []09:20
-k_matsuda:#shogun- individual_kernel = casted_kernel.get_first_kernel()09:20
-k_matsuda:#shogun- while individual_kernel09:20
-k_matsuda:#shogun- w << individual_kernel.get_combined_kernel_weight()09:20
-k_matsuda:#shogun- individual_kernel = casted_kernel.get_next_kernel()09:20
-k_matsuda:#shogun- end09:20
-k_matsuda:#shogun- p w09:20
k_matsudait seems work well...09:20
sonne|workk_matsuda: I think there is even a get_subkernel_weights() function so you don't need the loop09:34
k_matsudawhat is as argument for get_subkernel_weights() ?09:41
k_matsudai tryed that :09:41
-k_matsuda:#shogun- casted_kernel.get_subkernel_weights(casted_kernel.get_num_subkernels)09:43
k_matsudabut i get :09:43
-k_matsuda:#shogun- Wrong arguments for overloaded method 'CombinedKernel.get_subkernel_weights'. (ArgumentError)09:43
-k_matsuda:#shogun- Possible C/C++ prototypes are:09:43
-k_matsuda:#shogun- void CombinedKernel.get_subkernel_weights(int32_t &num_weights)09:43
-k_matsuda:#shogun- void CombinedKernel.get_subkernel_weights(float64_t **weights, int32_t *num_weights09:43
k_matsudai'm not familiar to SWIG/Ruby.. i dont know how to reference or pointer for native code.09:44
sonne|workok that is a bug09:53
k_matsudaI want interface like : float64_t * CombinedKernel::get_subkernel_weight(void)09:59
k_matsudai use loop of get_first_kernel()/get_next_kernel() a while10:00
k_matsudavery thanks for advice!10:01
15SAAI18Mshogun: Soeren Sonnenburg master * r50fb405 / (2 files): add get/set functions for combined kernel weights (forgotten sgvector transition) - http://git.io/RZrbXA10:08
sonne|workk_matsuda: ok submitted10:08
shogun-buildbotbuild #384 of cmdline_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/384  blamelist: sonne@debian.org10:11
shogun-buildbotbuild #359 of python_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/359  blamelist: sonne@debian.org10:13
k_matsudaI leave once, but I'll continue check channel logs, and greatly appreciate every advices.10:40
-!- k_matsuda [~k_matsuda@peter.pi.titech.ac.jp] has quit [Remote host closed the connection]10:41
-!- sonne|work [~sonnenbu@194.78.35.195] has left #shogun []14:27
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun14:28
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun14:52
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Ping timeout: 265 seconds]15:57
-!- sonne|work [~sonnenbu@194.78.35.195] has quit [Quit: Leaving.]16:20
15SAAI18Mshogun: Soeren Sonnenburg master * r2b03faf / src/configure :17:42
15SAAI18Mshogun: disable hmm parallel learning/structures and thus parallel hmm17:42
15SAAI18Mshogun: training/application at all.17:42
15SAAI18Mshogun: this is a temporary workaround to not see crashes and wrong results when17:42
15SAAI18Mshogun: learning - http://git.io/nrr5rw17:42
15SAAI18Mshogun: Soeren Sonnenburg master * r1a2c5b0 / examples/undocumented/libshogun/classifier_mklmulticlass.cpp : adjust example to SGVector style - http://git.io/9jsrVg19:56
15SAAI18Mshogun: Soeren Sonnenburg master * r2c4ee34 / (9 files in 6 dirs): fix other places to work with SGVector - http://git.io/kUNIhQ19:56
shogun-buildbotbuild #386 of cmdline_static is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/38620:01
shogun-buildbotbuild #387 of cmdline_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/387  blamelist: sonne@debian.org20:06
shogun-buildbotbuild #361 of python_static is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/36120:08
shogun-buildbotbuild #362 of python_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/362  blamelist: sonne@debian.org20:08
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun21:46
-!- in3xes [~in3xes@180.149.49.227] has quit [Read error: Connection reset by peer]21:48
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun22:05
-!- in3xes [~in3xes@180.149.49.227] has quit [Read error: Connection reset by peer]22:05
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun22:27
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun22:35
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Remote host closed the connection]22:35
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 248 seconds]22:36
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun22:39
-!- in3xes [~in3xes@180.149.49.227] has quit [Remote host closed the connection]22:43
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun22:44
-!- in3xes [~in3xes@180.149.49.227] has quit [Remote host closed the connection]22:49
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun22:49
-!- blackburn [~blackburn@188.168.2.160] has joined #shogun22:57
blackburnsonney2k: forgot to join channel :)22:59
blackburnI guess you are sleeping already22:59
--- Log closed Thu Dec 08 00:00:19 2011

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!