IRC logs of #shogun for Friday, 2016-07-01

« Thursday, 2016-06-30 Index Saturday, 2016-07-02 »
--- Log opened Fri Jul 01 00:00:03 2016
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]00:22
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun00:54
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]01:04
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun01:37
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]01:54
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun01:54
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]02:29
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting]02:33
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun02:33
shogun-buildbotbuild #60 of xenial - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/60  blamelist: lambday <heavensdevil6909@gmail.com>02:42
-!- GandalfTheWizard [~Eva@112.10.170.115] has joined #shogun03:28
-!- GandalfTheWizard [~Eva@112.10.170.115] has quit [Read error: Connection timed out]03:44
shogun-buildbotbuild #22 of clang - thread analysis is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang%20-%20thread%20analysis/builds/22  blamelist: lambday <heavensdevil6909@gmail.com>, Viktor Gal <viktor.gal@maeth.com>03:45
shogun-buildbotbuild #20 of clang - undefined behaviour analysis is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang%20-%20undefined%20behaviour%20analysis/builds/20  blamelist: lambday <heavensdevil6909@gmail.com>, Viktor Gal <viktor.gal@maeth.com>03:48
shogun-buildbotbuild #33 of memleak - valgrind is complete: Failure [failed memory check]  Build details are at http://buildbot.shogun-toolbox.org/builders/memleak%20-%20valgrind/builds/33  blamelist: lambday <heavensdevil6909@gmail.com>, Viktor Gal <viktor.gal@maeth.com>05:36
shogun-buildbotbuild #1033 of nightly_none is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_none/builds/1033  blamelist: lambday <heavensdevil6909@gmail.com>, Viktor Gal <viktor.gal@maeth.com>05:47
-!- OXPHOS [171b5289@gateway/web/freenode/ip.23.27.82.137] has joined #shogun05:54
-!- OXPHOS1 [~OXPHOS@23.27.82.137] has joined #shogun06:02
-!- OXPHOS [171b5289@gateway/web/freenode/ip.23.27.82.137] has quit [Quit: Page closed]06:02
-!- OXPHOS1 [~OXPHOS@23.27.82.137] has quit [Remote host closed the connection]06:02
-!- OXPHOS [~OXPHOS@23.27.82.137] has joined #shogun06:03
-!- OXPHOS [~OXPHOS@23.27.82.137] has quit [Client Quit]06:03
-!- OXPHOS [~OXPHOS@23.27.82.137] has joined #shogun06:03
shogun-buildbotbuild #1163 of nightly_default is complete: Failure [failed test notebooks]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1163  blamelist: lambday <heavensdevil6909@gmail.com>, Viktor Gal <viktor.gal@maeth.com>06:25
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun06:44
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]06:51
-!- GandalfTheWizard [~Eva@112.10.170.115] has joined #shogun06:51
-!- OXPHOS [~OXPHOS@23.27.82.137] has quit [Ping timeout: 240 seconds]07:10
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun07:15
-!- GandalfTheWizard [~Eva@112.10.170.115] has quit [Quit: Leaving.]07:25
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]07:32
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun07:39
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Client Quit]07:42
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has joined #shogun07:44
c4goldswwiking Earth to Viktor07:50
-!- c4goldsw [4086df9d@gateway/web/cgi-irc/kiwiirc.com/ip.64.134.223.157] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]08:00
-!- sanuj [~sanuj@117.203.5.236] has joined #shogun11:46
-!- sanuj [~sanuj@117.203.5.236] has quit [Quit: Leaving]12:00
-!- HeikoS [~heiko@nat-192-143.internal.eduroam.ucl.ac.uk] has joined #shogun12:05
-!- mode/#shogun [+o HeikoS] by ChanServ12:05
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has joined #shogun12:12
-!- mode/#shogun [+o lambday] by ChanServ12:12
-!- sanuj [~sanuj@117.203.5.236] has joined #shogun12:37
sanujHeikoS, yo12:47
sanujlambday, yo12:48
@lambdaysanuj: yolo!12:50
sanuj:D12:50
sanujhow's life12:50
@lambdaysanuj: that's a NP-complete question!12:50
sanujlambday, busy?12:50
sanuj:D12:51
@lambdayask something easier :D12:51
sanujlambday, here is one, suggest a method to generate multi-dimensional regression data :D12:51
sanujnon-linear12:52
sanuji want to put it in shogun-data12:52
@lambdayerm.. dosen't mnist have anything for this already?12:52
sanujmnist is number classification right?12:54
sanujfrom images12:54
sanujlambday, i want to put something in toy for neural net regression cookbook12:54
@HeikoSsanuj: jojo12:57
@HeikoSsorry was busy yesterday12:57
@lambdaysanuj: well, you can use some well known datasets, or design your non-linear function to generate synthetic data12:57
@HeikoShow are things?12:57
sanujgood12:57
sanujHeikoS, still in france?12:57
@HeikoSsanuj: nope12:57
@HeikoSback out of EU12:57
@HeikoS :D12:57
sanujHeikoS, got time to review PRs :D12:58
sanujcool12:58
@HeikoSsanuj: yep am on it12:58
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun12:58
shogun-notifier-shogun-data: OXPHOS :master * a7ac1dc / testsuite/meta/multiclass_classifier/linear_discriminant_analysis.dat: https://github.com/shogun-toolbox/shogun-data/commit/a7ac1dce0f41b6cd2bd3e5094589f62738269cb512:58
shogun-notifier-shogun-data: multiclass lda integration test data12:58
shogun-notifier-shogun-data: Heiko Strathmann :master * 4b27a7a / testsuite/meta/multiclass_classifier/linear_discriminant_analysis.dat: https://github.com/shogun-toolbox/shogun-data/commit/4b27a7a5c8b99d66829691f750cf9bbfa4e649e512:58
shogun-notifier-shogun-data: Merge pull request #112 from OXPHOS/master12:58
shogun-notifier-shogun-data:12:58
shogun-notifier-shogun-data: multiclass_lda integration test dataset12:58
shogun-notifier-shogun: Saurabh7 :develop * 3720efe / src/shogun/multiclass/tree/CARTree.cpp: https://github.com/shogun-toolbox/shogun/commit/3720efe80725a8381eb2144b7bc3af4d35d1125012:59
shogun-notifier-shogun: remove unnecessary memcpy12:59
shogun-notifier-shogun: Heiko Strathmann :develop * 5570fa7 / src/shogun/multiclass/tree/CARTree.cpp: https://github.com/shogun-toolbox/shogun/commit/5570fa76922b4dd11c0fa2699dc0bf996fedb53112:59
shogun-notifier-shogun: Merge pull request #3327 from Saurabh7/cartmin12:59
shogun-notifier-shogun:12:59
shogun-notifier-shogun: Minor changes CART12:59
sanujlambday, what about this => input (a,b,c,d) output (a+b+c+d)^213:00
sanuji can't see the graph with multiple dimension13:00
@lambdaysanuj: yeah maybe add some noise also13:01
sanujokay, HeikoS are you also fine with this? ^ It's a dataset that i need to add in toy for neural net regression cookbook13:02
sanujlambday, how's internship?13:03
sanujwhat plans after this13:03
sanuji forgot what you were working on....13:03
@HeikoSsanuj: sorry13:06
@HeikoSdidnt follow13:06
@HeikoSwhat do you need?13:06
@HeikoSnew dataset?13:06
@HeikoSfor meta example?13:06
sanujHeikoS, yes13:07
@HeikoSsanuj: what do you need?13:07
sanujfor neural net regression cookbook13:07
sanuja multidimensional regression dataset13:07
@HeikoSwhy multi d?13:07
sanujwe only have 1d datasets in toy13:07
sanujHeikoS, you want to give 1D data to a neural net?13:07
@HeikoSsanuj: whats the difference to giving 1d data to krr or gp?13:07
@HeikoSor a rf?13:08
sanujHeikoS, it will have 1 node in the input layer13:08
@HeikoSyep13:08
sanujthen a hidden layer13:08
sanujand one node in output layer13:08
@HeikoSI see13:08
sanujHeikoS, is this okay? ^13:08
@HeikoSyes it is ok, but I see your point13:09
@HeikoSit is kinda bad for ullstrating how to build nns13:09
@HeikoSso yeah, we can add another dimension13:09
sanujokay13:09
@HeikoSI suggest you sample from a GP13:09
@HeikoSknow how to do that?13:09
sanujno13:09
sanuji'll google13:09
sanujso 2-D data?13:09
@HeikoS2d we can still plot13:10
@HeikoSis that ok for illstration?13:10
@HeikoSfor need even more?13:10
@HeikoSI mean in the cookbook13:10
sanuji think it's fine13:10
@HeikoSso 2d doesnt really seem very different to 1d to me :)13:10
@HeikoSsanuj: mmmh13:10
sanujwe can have less neurons in the hidden layer13:10
sanujhaha, yeah13:10
@HeikoSI suggest this:13:10
@HeikoStake the 1d dataset we  have13:10
shogun-buildbotbuild #61 of xenial - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/61  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>13:10
@HeikoSand just add random noise in all other dimensions13:11
@HeikoSindependent of labels13:11
@HeikoSso then the NN has to figure that out13:11
@HeikoSand you can add like 10 dimensions of noise13:11
@HeikoSis that good?13:11
sanujalright13:11
sanujthis will be 10-D regression dataset13:11
@HeikoSyep13:12
@HeikoSthink of some clever name13:12
@HeikoSit is just the old dataset with added noise13:12
sanujHeikoS, regression_10d_noisy_labels_train.dat13:12
@HeikoSsanuj: actually, maybe just say 10d13:13
@HeikoSwithout the "noisy"13:13
shogun-buildbotbuild #2921 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2921  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>13:13
@HeikoSdont generate new labels13:13
@HeikoSah no13:13
sanujregression_10d_noisy_features_train.dat13:13
@HeikoSso labels are shared13:13
sanujoh isee13:13
@HeikoScall it13:13
@HeikoS<old_feats_name>_with_9d_noise.dat13:14
-!- Saurabh7_ [~Saurabh7@1.39.98.48] has joined #shogun13:14
@HeikoSand re-use the labels13:14
@HeikoSSaurabh7_: jo!13:14
sanujcool13:14
Saurabh7_HeikoS, hi13:14
@HeikoSSaurabh7_: just reading you pr13:14
Saurabh7_HeikoS, so  few things about the current PR13:14
@HeikoSyeah, not sure I get everything13:14
@HeikoScan we discuss here?13:14
Saurabh7_HeikoS, https://github.com/shogun-toolbox/shogun/blob/develop/tests/unit/multiclass/BaggingMachine_unittest.cc#L3513:15
Saurabh7_this test13:15
@HeikoSlooks ugly ;)13:15
Saurabh7_I havent implemented the shallow_copy_subset for labels base calss13:15
Saurabh7_so its not working13:15
Saurabh7_since it uses the mocklabels class13:15
@HeikoSSaurabh7_: wait but the method should be in base class SG_NOTIMPLEMENTED right?13:16
@HeikoSand the subclasses override?13:16
@HeikoSthat should be possible to fix no?13:16
Saurabh7_HeikoS, yes13:16
@HeikoScool13:16
Saurabh7_HeikoS,  yes possible to fix bt want to know how13:16
Saurabh7_HeikoS, should i jsut add it to the mock labels file13:16
Saurabh7_?13:16
Saurabh7_HeikoS, https://github.com/shogun-toolbox/shogun/blob/develop/tests/unit/labels/MockLabels.h woudl that be good13:17
@HeikoSSaurabh7_: maybe yes13:17
@HeikoSbut reuse your code13:17
@HeikoSsomehow wrap the call13:17
Saurabh7_HeikoS, ok I will do that then13:18
shogun-notifier-shogun: Sanuj :develop * a1d607a / / (4 files): https://github.com/shogun-toolbox/shogun/commit/a1d607acd7d7e3437a169f754689a03d2bc464d213:18
shogun-notifier-shogun: add cookbook for mkl regression13:18
shogun-notifier-shogun: Heiko Strathmann :develop * 738796d / / (4 files): https://github.com/shogun-toolbox/shogun/commit/738796d02af5474483f7ae5827dba3df8dc274e213:18
shogun-notifier-shogun: Merge pull request #3299 from sanuj/mkl_regression13:18
shogun-notifier-shogun:13:18
shogun-notifier-shogun: add cookbook for mkl regression13:18
Saurabh7_HeikoS,  also this shallow subset copy is going to be only thread safe for 'read' in current state13:19
Saurabh7_I mean if we change using say 'set_feature_vector' the original changes13:20
shogun-notifier-shogun: Sanuj :develop * 0030a87 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/0030a87a7f9442fea4da6a0ec8dc2e1c82f8532a13:21
shogun-notifier-shogun: add cookbook for neural net classification13:21
shogun-notifier-shogun: Heiko Strathmann :develop * 0939e27 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/0939e273dd585afd0c3b91b5585c34422fa5b95a13:21
shogun-notifier-shogun: Merge pull request #3246 from sanuj/cookbook_nn13:21
shogun-notifier-shogun:13:21
shogun-notifier-shogun: add cookbook for neural net classification13:21
@HeikoSSaurabh7_: yep I know13:22
@HeikoSSaurabh7_: we need to work on this13:22
@HeikoSthe interface that is returned from this shallow subset thing should only contain const methods in fact13:22
@HeikoSOr Sergey suggested a thing:13:22
@HeikoSonce the data is modified, it will copy the data and modify that13:22
@HeikoSbut thats kinda implicit and i dont like implicit things13:23
@HeikoSbut adding a read-only interface for features might be annoying as well13:23
@HeikoSThough it is possible13:23
@HeikoSwe have a ConstFeatures base class that has the same interface as Features13:23
@HeikoSand then we just wrap all the const methods from Features13:23
@HeikoSand done13:23
@HeikoSSee what I mean?13:23
Saurabh7_for algos like rf it might be okay since we dont do write13:23
@HeikoSYou could try that for dense features in a new pr and then we can discuss there13:23
@HeikoSI think that would be good13:23
@HeikoSin partoicular for the x-validation13:24
@HeikoSwhere we rely on the fact that machines dont change data13:24
@HeikoSin fact13:24
Saurabh7_constFeatures derived from Cfeature ?13:24
@HeikoSit should be the default that algorithms cannot change the features13:24
@HeikoSSaurabh7_:  let me think about this a bit more and then get back to you13:24
@HeikoSyou can also think13:24
@HeikoSSaurabh7_: for now, we just are careful that we dont change things13:24
@HeikoSSaurabh7_:  once the parallel rf is working, we can do re-use feature data in xvalidation13:25
Saurabh7_ok13:25
Saurabh7_also for setting random seed, when I do sg_rand->set_seed13:26
Saurabh7_and run multicore rf13:26
Saurabh7_the indices keep changin13:26
Saurabh7_HeikoS, as in the seed is not working13:26
@HeikoSwhich argument do you pass?13:27
@HeikoSto set_seed?13:27
Saurabh7_set_seed(1)13:27
@HeikoSmmh13:27
Saurabh7_in unittest13:27
@HeikoSmaybe the openmp randomizses the scheduling?13:27
Saurabh7_I found a issue like this which was merged13:27
Saurabh7_https://github.com/shogun-toolbox/shogun/pull/142413:27
@HeikoSmmmh13:28
@HeikoSyeah I see13:28
@HeikoSso how to solve?13:28
sanujHeikoS, why is that error in svm multiclass cookbook?13:28
Saurabh7_I tired setting m_rng like this for RF too but it didnt work13:29
@HeikoSsanuj: error?13:33
@HeikoSSaurabh7_: mmh13:34
@HeikoSSaurabh7_: sure you didnt use the static object?13:34
sanujHeikoS, https://gist.github.com/sanuj/1c1f0393f96e46c4b822057fabe9bb5f13:34
@HeikoSah13:34
@HeikoSthat is a parser error13:35
@HeikoSweird13:35
@HeikoSI dont know13:35
sanujokay13:35
@HeikoSseems like a syntax error in meta language13:35
sanujHeikoS, can you also review tags PR13:36
B-ZDM-Gok I am kinda back13:36
-!- B-ZDM-G is now known as bzdmgz13:36
sanujbzdmgz, !!!!!!!!!!13:37
bzdmgzwhat's up13:37
sanuji think tags PR will be merged soon13:37
sanujthen will start Tags swig again13:37
sanujbzdmgz, should start work on plugins from this weekend13:37
@HeikoSbzdmgz: bzdmgz!!!!13:38
@HeikoSwelcome13:38
@HeikoSsanuj: will do now13:38
Saurabh7_HeikoS, not sure, I did m_rng =sg_rand just like in xval13:38
@HeikoSsanuj, Saurabh7_ see https://github.com/shogun-toolbox/shogun/issues/332813:38
sanujHeikoS, https://github.com/shogun-toolbox/shogun/pull/322113:38
@HeikoSSaurabh7_: thats static I think13:39
@HeikoSSaurabh7_: mmh13:39
bzdmgzhttp://www.mediapost.com/publications/article/279260/yandex-hyper-targeted-ad-platform-uses-emails-tel.html13:39
@HeikoShaving multiple rngs is not good13:39
bzdmgzthat's what I was so busy about13:39
bzdmgz:D13:39
@HeikoSI see13:39
@HeikoSnice one13:39
@HeikoS!13:39
@HeikoSI like that there is donald trup coverage at the bottom of the article :)13:39
Saurabh7_so diffrent CRandom object for each tree ?13:40
Saurabh7_davai bzdmgz !13:40
@HeikoSSaurabh7_: yeah but we dont want that13:40
@HeikoSSaurabh7_: I think you can fix this via telling openmp to do static scheduling13:40
@HeikoSright?13:40
@HeikoSor no13:40
@HeikoSactually no13:40
@HeikoSit is weitrd13:40
@HeikoSthe problem is caused since all trees access the rng13:40
@HeikoSin parallel13:40
@HeikoSso the order is undefined13:40
@HeikoSit is thread safe13:41
@HeikoSbut one cannot guarantee fixed results13:41
Saurabh7_ye if it affects only testing its fine13:41
@HeikoSone way to overcome is that each tree has its own rng13:41
@HeikoSand usually, this one is shared13:41
@HeikoSbut for the test, you can give them each their own with a fixed seed13:41
@HeikoSwould that do it?13:41
Saurabh7_maybe I will check13:42
@HeikoSsanuj: reviewing the pr now, putting some comments13:45
@HeikoSif you reply straight away we might be able to merge soon13:45
sanujokay13:45
sanujHeikoS, _* are private methods13:48
@HeikoSsanuj: yep just figured13:48
@HeikoSstill13:48
@HeikoSbetter names pls13:48
@HeikoSthe name should indicate what the method does13:48
sanujHeikoS, okay13:48
@HeikoSmmmmh13:49
@HeikoSwill you use rvalue as well for these methods?13:49
@HeikoS_parameters() = food?13:49
@HeikoSfoo13:49
sanujrvalue?13:49
@HeikoSok nevermind13:49
@HeikoSthen rename pls13:49
sanujyeah13:49
sanujwill read about rvalue also13:50
sanujHeikoS, std::vector is used by all_parameters()13:51
@HeikoSyes I know13:52
sanuji tried to use SGVector13:52
@HeikoSbut <vector> cannot be included in CSGObject.h13:52
@HeikoSabsolutely no13:52
sanujbut it was creating complications in SWIG13:52
@HeikoSthink what happens if you do that13:52
sanujHeikoS, what?13:52
@HeikoSit is the base class13:52
@HeikoSso vector now is included everywhere in shogun13:53
sanujyeah13:53
shogun-buildbotbuild #8 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/8  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>13:53
sanujHeikoS, if I include SGVector, then that will also be included everywhere13:53
@HeikoScant you just forward declare the vector?13:53
@HeikoSand then include in the cpp13:53
@HeikoSthat would be ok13:53
sanujalright13:53
sanujyes13:53
@HeikoSknow what forward delcaring is?13:53
sanujyeah13:54
@HeikoScool13:54
sanuj:D13:54
shogun-buildbotbuild #2922 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2922  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>13:56
shogun-buildbotbuild #62 of xenial - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/62  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>13:58
sanujHeikoS, get() is not returning an empty any object14:04
@HeikoSok14:09
@HeikoSsorry14:09
-!- travis-ci [~travis-ci@ec2-54-227-50-239.compute-1.amazonaws.com] has joined #shogun14:12
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/14158872714:12
-!- travis-ci [~travis-ci@ec2-54-227-50-239.compute-1.amazonaws.com] has left #shogun []14:12
-!- Saurabh7_ [~Saurabh7@1.39.98.48] has quit [Ping timeout: 260 seconds]14:13
sanujHeikoS, how else to #include <../tests/unit/base/MockObject.h>14:15
sanujwithout relative paths14:15
@HeikoSah yes14:16
@HeikoSsorry :)14:16
sanujHeikoS, if a class uses new parameter framework, will it require to have member variables?14:18
sanujsince the variables are being stored in a map now14:18
@HeikoSsanuj: nono14:22
@HeikoSbut we want to checl/support both cases14:22
@HeikoSif we re-work the SG_ADD thing14:23
@HeikoSwe register existing members14:23
sanujalright14:23
@HeikoSin the future, we might refactor that so that there is no member14:23
@HeikoSbut allowing this makes refactoring the old code easier14:23
@HeikoSyou see?14:23
sanujyes14:23
@HeikoSsince we dont need to touch all shognu classes and delete the members14:23
@HeikoSalso, it gives super fast access to members inside a class, which is also kinda nice14:23
@HeikoSsince get() might be slower14:23
bzdmgzuhm14:23
bzdmgzis it all working now?14:24
@HeikoSbzdmgz: yes more or less14:24
sanujHeikoS, different names can have same hash which might result in collision14:24
@HeikoSbzdmgz: just did a lot of comments, but these are all minor and style14:24
@HeikoSsanuj: ok, then leave the strong comparison in14:24
@HeikoSsanuj: this is merge ready once the comments are addressed14:25
@HeikoSsanuj: think about which parts are exposed to shogun users14:25
@HeikoSand document them in a way that is non-confusing14:25
sanujHeikoS, thanks for the comments, it took lot of your time :)14:25
sanujokay14:25
@HeikoSsanuj: this is looking great otherwise14:25
sanujcool14:25
sanujshogun-buildbot, dance14:25
@HeikoSso we can merge later today no?14:25
shogun-buildbot<(^.^<)14:25
shogun-buildbot<(^.^)>14:25
shogun-buildbot(>^.^)>14:25
shogun-buildbot(7^.^)714:25
shogun-buildbot(>^.^<)14:25
@HeikoSbzdmgz: dance14:25
sanujHeikoS, yes, i'll update this now14:26
@HeikoSbzdmgz: shall we merge to feature branch14:27
@HeikoSor develop?14:27
sanujfeature14:27
@HeikoSbzdmgz: it only adds things14:27
bzdmgzfeature14:27
sanujtest on buildbot14:27
bzdmgzI think feature14:27
@HeikoSah yeah14:27
@HeikoSok14:27
bzdmgzis it swig ready?14:27
@HeikoSsanuj: make sure to rebase against develop then every now and then14:27
bzdmgzif yes, then develop14:27
@HeikoSbzdmgz: after merge, the swig parts come14:27
@HeikoSokok14:27
sanujwe have 165 conversations in this PR14:28
@HeikoSgetting lunch see you later14:28
@HeikoSsanuj: yeah need to merge it14:28
sanujHeikoS, enjoy your lunch14:28
@HeikoSand then start a new one14:28
@HeikoSthanks14:28
sanujbzdmgz, good to have you back :D14:29
bzdmgzoh well I am partially back14:29
bzdmgzand also I am out to mallorca in 2 weeks14:29
-!- Saurabh7_ [~Saurabh7@59.88.89.134] has joined #shogun14:32
sanujlol14:34
sanujbzdmgz, spain?14:35
shogun-buildbotbuild #9 of FC23 - libshogun - aarch64 is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/9  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Sanuj <sanuj.sharma.in@gmail.com>14:35
bzdmgzsanuj: yeap14:35
shogun-notifier-shogun: lambday :feature/bigtest * 725d036 / src/shogun/statistics/experimental/internals/ (34 files): https://github.com/shogun-toolbox/shogun/commit/725d036df296af116207ed17a9fa895c17fa812615:10
shogun-notifier-shogun: added first version of bigtesting framework15:10
shogun-notifier-shogun: lambday :feature/bigtest * f2a6d89 / src/shogun/statistics/experimental/internals/ (5 files): https://github.com/shogun-toolbox/shogun/commit/f2a6d89ec7b5f0699fd71ca030994155b153fd3815:10
shogun-notifier-shogun: documentation and code-style changes15:10
shogun-notifier-shogun: lambday :feature/bigtest * 11fbddd / src/shogun/statistics/experimental/ (3 files): https://github.com/shogun-toolbox/shogun/commit/11fbddd6bd04aac981af30327738b2bf681f804315:10
shogun-notifier-shogun: added hypothesis test (experimental) base class15:10
shogun-notifier-shogun: lambday :feature/bigtest * 391f644 / src/shogun/ (108 files): https://github.com/shogun-toolbox/shogun/commit/391f6448d17f22fc5788c98cad1d09ea5fe1fb6515:10
shogun-notifier-shogun: removed old files15:10
shogun-notifier-shogun: lambday :feature/bigtest * a611c66 / src/shogun/hypothesistest/ (38 files): https://github.com/shogun-toolbox/shogun/commit/a611c66bbbf99771efc3f9031124514f0c9e4a8f15:10
shogun-notifier-shogun: added rest of the files15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * ada18dc / examples/meta/src/statistical_testing/quadratic_time_mmd.sg: https://github.com/shogun-toolbox/shogun/commit/ada18dc05879d27216902b82d7ec2f5122795a3115:10
shogun-notifier-shogun: first meta example for mmd15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 16f8ae9 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/16f8ae9e3163a205c9d073282426a7418c88597615:10
shogun-notifier-shogun: Add an initial cookbook page for quadratic time mmd15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 75afdd2 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/75afdd2a40b77b602517586100f8f48bcb73f52615:10
shogun-notifier-shogun: add linear time mmd api example draft15:10
shogun-notifier-shogun: lambday :feature/bigtest * 74bfd50 / / (127 files): https://github.com/shogun-toolbox/shogun/commit/74bfd505fd7011f38896373898c0ec80b209fe0715:10
shogun-notifier-shogun: fix build issues15:10
shogun-notifier-shogun: temporarily removed unit-tests to avoid compilation issues15:10
shogun-notifier-shogun: lambday :feature/bigtest * 4db7d60 / src/shogun/statistical_testing/MMD.cpp,src/shogun/statistical_testing/MMD.h: https://github.com/shogun-toolbox/shogun/commit/4db7d60dda9878829871419947cbae20cd7debfe15:10
shogun-notifier-shogun: removed statistic computation API for multiple kernels15:10
shogun-notifier-shogun: lambday :feature/bigtest * 93fcb56 / src/shogun/statistical_testing/MMD.cpp: https://github.com/shogun-toolbox/shogun/commit/93fcb5605447f19ad6932b7d725edee9866d604515:10
shogun-notifier-shogun: removed simulate_null15:10
shogun-notifier-shogun: lambday :feature/bigtest * d920247 / src/shogun/statistical_testing/MMD.cpp: https://github.com/shogun-toolbox/shogun/commit/d92024765c4fe738094f58260f03ecbcf60e74d515:10
shogun-notifier-shogun: added permutation test with precomputed kernel matrices15:10
shogun-notifier-shogun: lambday :feature/bigtest * 49b0fdb / examples/meta/src/statistical_testing/linear_time_mmd.sg: https://github.com/shogun-toolbox/shogun/commit/49b0fdb1aafbd3fe48d31a9bbffbbb8be13ec0bd15:10
shogun-notifier-shogun: added number of samples (mandatory for streaming samples) to lineartime mmd meta example15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * cbce1c3 / src/shogun/statistical_testing/QuadraticTimeMMD.cpp: https://github.com/shogun-toolbox/shogun/commit/cbce1c34e50a3fcc41d6679de354da236ab0d9d415:10
shogun-notifier-shogun: todo for later15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 37e0d41 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/37e0d41de0e54c6d4571fcdb635c04ddbd1195c215:10
shogun-notifier-shogun: add constructor to QuadraticTimeMMD15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * e732772 / doc/cookbook/source/examples/statistical_testing/quadratic_time_mmd.rst: https://github.com/shogun-toolbox/shogun/commit/e732772634e059212a5bcef955816052cd8ccc2515:10
shogun-notifier-shogun: Give the example some love15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 0258a3f / doc/cookbook/source/examples/statistical_testing/ (2 files): https://github.com/shogun-toolbox/shogun/commit/0258a3f2e6cbc695ab651558ed880c79359148a915:10
shogun-notifier-shogun: more cookbook updates15:10
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * d258d01 / src/shogun/statistical_testing/QuadraticTimeMMD.cpp: https://github.com/shogun-toolbox/shogun/commit/d258d01c4e767648ca17acbe2bc4096f5df613bc15:10
shogun-notifier-shogun: added data-fetchers and kernel-manager unit-tests15:11
shogun-notifier-shogun: lambday :feature/bigtest * 90c6174 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/90c6174a24b1542d8348c67c4aae0fcae409a91315:11
shogun-notifier-shogun: fixed quadratic time mmd bugs and speedup15:11
shogun-notifier-shogun: lambday :feature/bigtest * 023f9fb / / (3 files): https://github.com/shogun-toolbox/shogun/commit/023f9fbcda54c088dfc25870894a1baec2d26ea915:11
shogun-notifier-shogun: added more tests, fixed more bugs15:11
shogun-notifier-shogun: lambday :feature/bigtest * a8febee / / (6 files): https://github.com/shogun-toolbox/shogun/commit/a8febee5a07381a420517d935253a121501db3a915:11
shogun-notifier-shogun: removed temporary permuted matrix creation for mmd15:11
shogun-notifier-shogun:15:11
shogun-notifier-shogun:   - since SIMD doesn't help in this case, rather slows things down15:11
shogun-notifier-shogun:     due to temporary matrix creation15:11
shogun-notifier-shogun: lambday :feature/bigtest * df1a34d / / (5 files): https://github.com/shogun-toolbox/shogun/commit/df1a34db0c40e994dca7a1a2692a7429198fc2fd15:11
shogun-notifier-shogun: added cache-friendly sum-computation for permutation15:11
shogun-notifier-shogun: lambday :feature/bigtest * 86f68e6 / src/shogun/statistical_testing/ (4 files): https://github.com/shogun-toolbox/shogun/commit/86f68e6ae3d4187fb81303d7d79be4c8847f895415:11
shogun-notifier-shogun: Added task-parallel computation support15:11
shogun-notifier-shogun:15:11
shogun-notifier-shogun:   - Speeded up the permutation test15:11
shogun-notifier-shogun:   - Overall speed-up due to15:11
shogun-notifier-shogun:     (a) cache-friendly computation and15:11
shogun-notifier-shogun: lambday :feature/bigtest * 9a0e277 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/9a0e277a195044b0e6d7ada96259b764739a083f15:11
shogun-notifier-shogun: added feature shallow copy and merged copy methods15:11
shogun-notifier-shogun: lambday :feature/bigtest * e7d1b8e / / (2 files): https://github.com/shogun-toolbox/shogun/commit/e7d1b8ecb44c6b3186efb0ea2e5d679bcf07e4da15:11
shogun-notifier-shogun: made the data manager work with shallow copy15:11
shogun-notifier-shogun: lambday :feature/bigtest * 4547f20 / tests/unit/statistical_testing/ (2 files): https://github.com/shogun-toolbox/shogun/commit/4547f20fcac72532f1bc969676f05f86eff0bebd15:11
shogun-notifier-shogun: added kernel selection unit tests15:11
shogun-notifier-shogun: lambday :feature/bigtest * 3d91eff / / (3 files): https://github.com/shogun-toolbox/shogun/commit/3d91eff82824bf786a2eb244cc1e4d48be66205915:11
shogun-notifier-shogun: added block data structure15:11
shogun-notifier-shogun: lambday :feature/bigtest * 5ac309e / src/shogun/statistical_testing/internals/ (2 files): https://github.com/shogun-toolbox/shogun/commit/5ac309e4aa04b7606d004e9c882ffff5ba1f264a15:11
shogun-notifier-shogun: updated next samples data structure with blocks15:11
shogun-notifier-shogun: lambday :feature/bigtest * efc29dc / / (6 files): https://github.com/shogun-toolbox/shogun/commit/efc29dc3f6a96fc0ed0d9ccc6d1dbcf017c2791c15:11
shogun-notifier-shogun: updated data fetchers to return naked pointers15:11
shogun-notifier-shogun: lambday :feature/bigtest * 0ea3ded / src/shogun/statistical_testing/internals/DataManager.cpp: https://github.com/shogun-toolbox/shogun/commit/0ea3ded45da768b4d22ddc8b939d83a9e652ba5215:11
shogun-notifier-shogun: updated data manager to work with blocks15:11
shogun-notifier-shogun: lambday :feature/bigtest * 45b50ed / src/shogun/statistical_testing/HypothesisTest.cpp: https://github.com/shogun-toolbox/shogun/commit/45b50edfc82ebdc384af94461276c14e25e244c415:11
shogun-notifier-shogun: removed unused variable result15:11
shogun-notifier-shogun: lambday :feature/bigtest * 656b51e / src/shogun/statistical_testing/ (2 files): https://github.com/shogun-toolbox/shogun/commit/656b51ee6e497f88cfcdd546dcdf1fdc56a2e91415:11
shogun-notifier-shogun: added convenient constructor15:11
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 5283bcd / examples/meta/src/statistical_testing/ (2 files): https://github.com/shogun-toolbox/shogun/commit/5283bcde8c380d17778f34add3f10b2279208af715:11
shogun-notifier-shogun: added compiling (but not running) examples of kernel selection for MMD15:11
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * f7c93d3 / doc/cookbook/source/examples/statistical_testing/linear_time_mmd_kernel_selection.rst: https://github.com/shogun-toolbox/shogun/commit/f7c93d3077624e9a666259a266a4d40e0ded344815:11
shogun-notifier-shogun: initial sketch for kernel learning example15:11
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * e26a89d / doc/ipython-notebooks/ (2 files): https://github.com/shogun-toolbox/shogun/commit/e26a89d7df083cedadf147bef8e9bcd8f42b8a3c15:11
shogun-notifier-shogun: rename folder15:11
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * d6c4d04 / examples/meta/src/statistical_testing/quadratic_time_mmd_kernel_selection.sg: https://github.com/shogun-toolbox/shogun/commit/d6c4d0431ff3577c1d67f3eb486270ea7caef2d215:11
shogun-notifier-shogun: use different data in testing than in kernel learning15:11
shogun-notifier-shogun: lambday :feature/bigtest * ed316a4 / / (9 files): https://github.com/shogun-toolbox/shogun/commit/ed316a4c463b73e5bcda8027643df5f807dd6e8515:11
shogun-notifier-shogun: renamed OPTIMIZE_MMD to MAXIMIZE_POWER in kernel selection15:11
shogun-notifier-shogun: lambday :feature/bigtest * 1adc869 / src/shogun/statistical_testing/MMD.cpp: https://github.com/shogun-toolbox/shogun/commit/1adc869b23dfde6fad48d287e17e7cfb6c1a840515:11
shogun-notifier-shogun: removed precomputed kernels (if there are any) while selecting kernel15:11
shogun-notifier-shogun: lambday :feature/bigtest * 4d1d131 / src/shogun/statistical_testing/MMD.cpp: https://github.com/shogun-toolbox/shogun/commit/4d1d1316444967d691b47ad7f935db4fb63633c515:11
shogun-notifier-shogun: added statistic and Q computation methods15:11
shogun-notifier-shogun: lambday :feature/bigtest * d120e78 / src/shogun/statistical_testing/ (14 files): https://github.com/shogun-toolbox/shogun/commit/d120e78565b06e7c722d447169360a336bf3bc3215:11
shogun-notifier-shogun: added weighted kernel learning methods15:11
shogun-notifier-shogun: lambday :feature/bigtest * 466e915 / src/shogun/statistical_testing/internals/OptimizationSolver.cpp: https://github.com/shogun-toolbox/shogun/commit/466e915fabbe54dd72e2ec10de137f1cfc4c07dd15:11
shogun-notifier-shogun: temporary turned off GPL flag for QP solver15:11
shogun-notifier-shogun: lambday :feature/bigtest * 7738308 / src/shogun/statistical_testing/ (6 files): https://github.com/shogun-toolbox/shogun/commit/77383080f645088d6b09b3f50dc517dab1fb728f15:12
shogun-notifier-shogun: added support for cross-validation setting for kernel selection (incomplete)15:12
shogun-notifier-shogun: lambday :feature/bigtest * 52252d4 / / (20 files): https://github.com/shogun-toolbox/shogun/commit/52252d4a5e1d820e4aaedc3f96e077ab03eb597a15:12
shogun-notifier-shogun: made the enum classes in statistical testing simple enums15:12
shogun-notifier-shogun: lambday :feature/bigtest * e310a27 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/e310a2739975ddececc8ed45ab3d0ecfc6d185aa15:12
shogun-notifier-shogun: added first draft of train-test data split15:12
shogun-notifier-shogun: lambday :feature/bigtest * 5c35b41 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/5c35b41ced8859ca590c28308b188aa5d37d16b715:12
shogun-notifier-shogun: fixed clone subset stack bug in features util15:12
shogun-notifier-shogun: lambday :feature/bigtest * ea78548 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/ea785484e4ff35f984d4a1d67ca1f899e1e844e215:12
shogun-notifier-shogun: full and blockwise train/test data fetchers fixed15:12
shogun-notifier-shogun: lambday :feature/bigtest * c613b04 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/c613b048be8d6c6d9ddb7b87c98aeeb1f1914c4a15:12
shogun-notifier-shogun: fixed memory bug for streaming data parser and added unit-test15:12
shogun-notifier-shogun: lambday :feature/bigtest * 6beb360 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/6beb3604029e5a3954537f3aadb4adb6a3fbc1ed15:12
shogun-notifier-shogun: full and blockwise train/test streaming data fetchers added15:12
shogun-notifier-shogun: lambday :feature/bigtest * 1820556 / / (11 files): https://github.com/shogun-toolbox/shogun/commit/18205562c8e18cabf08facf2e689fd934d9cbaae15:12
shogun-notifier-shogun: intermediate work for test-train data split15:12
shogun-notifier-shogun: lambday :feature/bigtest * 0537b6c / / (6 files): https://github.com/shogun-toolbox/shogun/commit/0537b6c9692265761d69cea75f8d4d76dfec997015:12
shogun-notifier-shogun: added permutation without storing the kernel matrix15:12
shogun-notifier-shogun: lambday :feature/bigtest * 0b54a9c / src/shogun/statistical_testing/ (16 files): https://github.com/shogun-toolbox/shogun/commit/0b54a9caedbfb78b1f3e164865cca5f1fa12443715:12
shogun-notifier-shogun: save the kernel selection measures for later query15:12
shogun-notifier-shogun: lambday :feature/bigtest * ce52681 / src/shogun/statistical_testing/internals/ (6 files): https://github.com/shogun-toolbox/shogun/commit/ce52681992a1ce65118cad50afc40584ddbacb2215:12
shogun-notifier-shogun: refactored maximize cross validation kernel selection method15:12
shogun-notifier-shogun: lambday :feature/bigtest * cef312c / src/shogun/statistical_testing/ (6 files): https://github.com/shogun-toolbox/shogun/commit/cef312c1db9ac9dcb6dc430513e333c39797d19315:12
shogun-notifier-shogun: refactor API (incomplete)15:12
shogun-notifier-shogun: lambday :feature/bigtest * cfa8dcb / / (31 files): https://github.com/shogun-toolbox/shogun/commit/cfa8dcb6dd420be5725b2f14e9891df3da2a6a1f15:12
shogun-notifier-shogun: refactored train/test subsetting and hypothesis test framework15:12
shogun-notifier-shogun: lambday :feature/bigtest * 1ee330a / src/interfaces/modular/Statistics.i,src/interfaces/modular/Statistics_includes.i: https://github.com/shogun-toolbox/shogun/commit/1ee330a7d46945a465ed1e06818e0a31cf0d2f7d15:12
shogun-notifier-shogun: added kernel selection strategy in modular interface15:12
shogun-notifier-shogun: lambday :feature/bigtest * 6758320 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/6758320a3b78679ea3081b0b33edb3f1d968d0e615:12
shogun-notifier-shogun: refactored the interface for specifying kernel selection strategy15:12
shogun-notifier-shogun: lambday :feature/bigtest * b10a31d / src/shogun/statistical_testing/internals/MaxXValidation.cpp: https://github.com/shogun-toolbox/shogun/commit/b10a31d9ecbef34535f09515c9bcf0034330650515:12
shogun-notifier-shogun: fixed cross validation bug15:12
shogun-notifier-shogun: lambday :feature/bigtest * d9f39f7 / / (41 files): https://github.com/shogun-toolbox/shogun/commit/d9f39f72cefd032c99a1608a20c623e76fe7df5115:12
shogun-notifier-shogun: refactored directory structure15:12
shogun-notifier-shogun: lambday :feature/bigtest * 9a6f1bf / tests/unit/statistical_testing/ (5 files): https://github.com/shogun-toolbox/shogun/commit/9a6f1bf67492b4ddfc89c67cf2bb6c365e09521815:12
shogun-notifier-shogun: updated kernel selection unit-tests15:12
shogun-notifier-shogun: lambday :feature/bigtest * 06866ad / src/shogun/ (8 files): https://github.com/shogun-toolbox/shogun/commit/06866ada9ca0a832017c20d250cae15f8c359adf15:12
shogun-notifier-shogun: added multi kernel permutation test cross validation15:12
shogun-notifier-shogun: lambday :feature/bigtest * 4450b58 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/4450b58eb1194cc261a8efd5e134890dd4e33f4615:12
shogun-notifier-shogun: fixed alpha bug ;)15:12
-!- sanuj [~sanuj@117.203.5.236] has quit [Ping timeout: 260 seconds]15:25
-!- travis-ci [~travis-ci@ec2-54-227-50-239.compute-1.amazonaws.com] has joined #shogun15:45
travis-ciit's lambday's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/14161187915:45
-!- travis-ci [~travis-ci@ec2-54-227-50-239.compute-1.amazonaws.com] has left #shogun []15:45
-!- sanuj [~sanuj@117.203.5.236] has joined #shogun15:45
sanujbzdmgz, there?15:46
bzdmgzsanuj: yea16:30
sanujbzdmgz, https://github.com/shogun-toolbox/shogun/pull/3221#discussion-diff-69287780R2516:31
sanuji think forward declaring std::vector is a bad idea16:31
sanujgotta have dinner16:34
sanujwill be back in 30 mins16:34
-!- sanuj [~sanuj@117.203.5.236] has quit [Ping timeout: 244 seconds]16:38
-!- sanuj [~sanuj@117.203.5.236] has joined #shogun17:08
sanujbzdmgz, ping17:09
-!- GandalfTheWizard [~Eva@112.10.170.115] has joined #shogun17:18
-!- GandalfTheWizard [~Eva@112.10.170.115] has quit [Quit: Leaving.]17:33
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]18:12
@HeikoSbzdmgz: jojo18:34
@HeikoSsanuj: hi18:36
sanujHeikoS, hi18:36
sanujHeikoS, can't forward declare std::vector18:39
sanujhttp://stackoverflow.com/questions/307343/forward-declare-an-stl-container18:39
@HeikoSsanuj: yep18:48
@HeikoStrue18:48
@HeikoSsanuj: we need to find another way18:49
@HeikoSbecause we cannot have vector in interface18:49
@HeikoSthere is a few candidates18:49
@HeikoSDynamicArray<const char*> would work18:49
@HeikoSbut thats a bit messy18:49
@HeikoSSGStringList18:50
sanujHeikoS, we need to consider SWIG also18:50
@HeikoSnot sure if typemaps support that18:50
@HeikoSyeah18:50
@HeikoShow will Tags be passed through the swig interfaces?18:50
@HeikoSwill they be at all?18:50
@HeikoSor just strings?18:50
sanujtags will be passed18:50
@HeikoSwhats the typemap18:51
@HeikoSfor a tag?18:51
@HeikoSi.e. how would the python call look like?18:51
sanujTag<Kernel> will be TagKernel18:51
@HeikoSI see18:51
@HeikoSso a tag class for every base class shogun has18:51
sanujyes18:51
sanujit'll be clear once i send the Tags PR18:52
sanujhere is an old working example18:52
sanujhttps://gist.github.com/sanuj/e24d003b5af990fee1b772129ace75fa18:52
sanujbut it won't work with the new changes18:52
sanujbut gives you an idea about interface18:52
sanujHeikoS, i have renamed _get() to get_param_with_btag()18:53
sanujand so on18:53
sanujbecause the private methods take basetag as argument18:54
@HeikoSyes good18:56
@HeikoSthe name is ugly, but is only used inside tag framework18:56
bzdmgzok here18:56
@HeikoSso thats good18:56
@HeikoSbzdmgz: hey18:56
bzdmgzwhat18:56
@HeikoScan you help me with a cookbook thing?18:56
bzdmgzyes18:56
@HeikoSbzdmgz: I want that "show full listing" buttoin18:56
bzdmgzuhh it is not that easy18:56
@HeikoSbzdmgz: why?18:57
@HeikoScan link to something18:57
@HeikoSmmh18:58
@HeikoSrthough that doesnt give syntax highlighting18:58
@HeikoSmaybe as a first isntance?18:58
@HeikoSlink to file?18:58
@HeikoSsanuj: I suggest you use DynamicArray18:59
@HeikoSsanuj: but you know what18:59
@HeikoSwe can do that stuff later18:59
@HeikoSI think you can just remove the method for now19:00
@HeikoSit is just for users to find out what parameters there are19:00
@HeikoSnot important for this PR19:00
@HeikoSwe need to add that later, but ok for now to remove it19:00
sanujsounds cool19:00
@HeikoSI would do it after the swig is in19:00
sanujalright19:00
-!- GandalfTheWizard [~Emma@112.10.170.115] has joined #shogun19:06
sanujHeikoS, the point of BaseTag is to store the hash value19:07
sanujeven though it's private19:07
sanuji need to talk about it in docs19:08
sanujotherwise Tag would have sufficed19:08
@HeikoSsanuj: if the doc is intended to be read by a shogun user19:08
@HeikoSit should not tag about tags at all19:08
@HeikoSbut just "representing a parameter"19:08
@HeikoSsomething like this19:08
@HeikoSsanuj: and rather than talking about hashs, just say "unique identifier§"19:09
@HeikoSsanuj: users just get confused by tags, hashes, etc19:09
@HeikoSthey just want to get the parameter value19:10
@HeikoSand if they read the docs19:10
@HeikoSit should tell them how to do that19:10
@HeikoSnothhing more19:10
@HeikoSthe docs for devs for internal stuff can be more involved19:10
@HeikoSsanuj: these things are really important for such low level user exposed interfaces19:11
@HeikoSsanuj: we had a guy ranting about shogun's ugly interface on twitter a while ago19:11
@HeikoS ;)19:11
-!- GandalfTheWizard [~Emma@112.10.170.115] has quit [Quit: Leaving.]19:16
sanujHeikoS, sorry had gone away for a while19:20
sanujback now19:20
@HeikoSno worries19:20
-!- Saurabh7_ [~Saurabh7@59.88.89.134] has quit [Ping timeout: 276 seconds]19:21
sanujwe will make shogun great again19:21
sanujbtw if this new parameters framework is slower than the old one19:21
sanujwhat's the point of doing it19:21
sanujfor plugins?19:21
bzdmgzit is deffs slower19:23
bzdmgzit simply doesn't matter19:23
sanujbzdmgz, how is it better than the old one then?19:23
bzdmgzbut it reduces LoC and simplifies many things19:23
sanujloc?19:24
bzdmgzand makes it possible to write plugins19:24
bzdmgzwe won't have all these getters19:24
bzdmgzour classes are mostly cluttered19:24
bzdmgzwith them19:24
sanujohkayyy19:24
sanujhmmm19:24
sanujyeah, no more getters and setters anymore19:24
sanujbzdmgz, btw we can think about adding things like one-shot learning in shogun in the future19:25
sanujbut it requires neural nets19:25
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun19:27
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * 3b03d18 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/3b03d18f02e3fd07889d4f3c221b1f83c6af5dc519:27
shogun-notifier-shogun: updated meta examples19:27
shogun-notifier-shogun: Heiko Strathmann :feature/bigtest * ddb1c89 / src/shogun/statistical_testing/internals/DataManager.cpp: https://github.com/shogun-toolbox/shogun/commit/ddb1c897924214bafe9b42c4405b96518f58123019:27
shogun-notifier-shogun: mini clean for error message19:27
@HeikoS(18:19:47) sanuj: we will make shogun great again :D19:28
@HeikoSsanuj: it will make things so much nicer19:29
@HeikoSsanuj: not sure it will be that slower to be honest19:29
@HeikoSever tried to serialize a shogun class?19:29
sanujhehe, no19:29
@HeikoSthats pretty much the only place where we used the old parameter framework before19:29
@HeikoSsanuj: btw once we have swig merged19:30
@HeikoSwe can do a prototype of changing SG_ADD to the new tags19:30
@HeikoSand then remove all setters of class19:30
@HeikoSand getters19:30
@HeikoSand see how that goes19:30
sanujyes19:30
@HeikoSok I gotta go now19:30
sanujokay, bye19:30
@HeikoSsanuj: or is the thing merge ready already?19:30
sanujtake caer19:30
sanujno, one hour more19:31
sanuji'll update the PR19:31
@HeikoSokok19:31
@HeikoSmight look tonight19:31
@HeikoSif not tomorrow19:31
sanujokay19:31
sanujbzdmgz, can i remove self() from here https://github.com/shogun-toolbox/shogun/pull/3221/files#diff-9c3599c0d2090e493be261b079e9b63eR15319:37
sanujam not sure what effect it will have19:37
-!- HeikoS [~heiko@nat-192-143.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]19:40
-!- travis-ci [~travis-ci@ec2-54-227-50-239.compute-1.amazonaws.com] has joined #shogun19:58
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/14167773219:58
-!- travis-ci [~travis-ci@ec2-54-227-50-239.compute-1.amazonaws.com] has left #shogun []19:58
sanujbzdmgz, ping19:59
bzdmgzsanuj: why to remove this20:03
sanujheiko was saying20:04
sanujbzdmgz, can you also tell what does this comment mean? https://github.com/shogun-toolbox/shogun/pull/3221#discussion-diff-69291096R37520:04
bzdmgzI don't think he means you should remove this20:04
bzdmgzsanuj: well it means just to use20:05
bzdmgzget(Tag<float64_t>("int"));20:05
sanujohkay20:05
sanujthanks20:06
-!- sanuj [~sanuj@117.203.5.236] has quit [Quit: Leaving]20:21
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has quit [Quit: Page closed]20:39
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]22:28
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun23:50
-!- mode/#shogun [+o HeikoS] by ChanServ23:50
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit]23:55
--- Log closed Sat Jul 02 00:00:04 2016
« Thursday, 2016-06-30 Index Saturday, 2016-07-02 »

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