Open in new window / Try shogun cloud
--- Log opened Thu Mar 29 00:00:19 2012
OgreMagi@sonney2k: there?00:03
-!- OgreMagi [~vinaygarg@14.139.82.6] has left #shogun []00:18
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has quit [Remote host closed the connection]00:30
shogun-buildbotbuild #425 of octave_modular is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_modular/builds/42500:32
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has joined #shogun00:33
n4nd0sonney2k: I am commenting more the code now :)00:51
-!- flxb [~cronor@e178172254.adsl.alicedsl.de] has quit [Read error: Connection reset by peer]00:55
-!- flxb [~cronor@e178172254.adsl.alicedsl.de] has joined #shogun00:55
-!- ansh [b49531e5@gateway/web/freenode/ip.180.149.49.229] has joined #shogun01:20
-!- ansh [b49531e5@gateway/web/freenode/ip.180.149.49.229] has quit [Quit: Page closed]01:26
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 260 seconds]02:01
-!- ankit [18e9e06a@gateway/web/freenode/ip.24.233.224.106] has joined #shogun02:37
-!- ankit [18e9e06a@gateway/web/freenode/ip.24.233.224.106] has left #shogun []02:39
-!- trailblazer [18e9e06a@gateway/web/freenode/ip.24.233.224.106] has joined #shogun02:40
-!- flxb [~cronor@e178172254.adsl.alicedsl.de] has quit [Quit: flxb]02:43
-!- puffin444 [230bf329@gateway/web/freenode/ip.35.11.243.41] has joined #shogun03:02
puffin444hello03:02
trailblazerhello are any moderators online ?03:07
trailblazeri have a question about the GSOC 1203:08
puffin444hi trailblazer I don't know if there are moderators here03:08
puffin444I'm a student too03:08
trailblazerare u applying for GSOC too ?03:09
puffin444I'll see if I can put together an application03:09
puffin444so yes03:10
trailblazeryea i was going through it ,it says  under one section ---"Make yourself familiar with the github interface and start to interact with us by sending a (small) patch via a github pull request."03:10
trailblazeri am not really sure how can i contributr to it03:10
trailblazerany ideas??03:10
puffin444There are instructions how to do that on Github. You make an account on github and make your own personal fork.03:11
trailblazeri mean i have a girhub set up and everything03:11
trailblazerits this part i am wondering  about  ---"start to interact with us by sending a (small) patch via a github pull request."03:12
puffin444Oh yeah material for a patch. Good question. I am actually submitting some initial code for my proposed project.03:12
trailblazerdo we need to work on some of their bugs or something03:12
puffin444Do you have a specific project in mind?03:12
trailblazeryep03:12
trailblazeri think i am gonna for ---Implement multitask and domain adaptation algorithms03:12
puffin444So one mentor told me any bug fix, even if it was unrelated to the proposed project would be helpful03:12
trailblazeri see03:13
puffin444You could also talk to the proposed mentor and ask if there is anything initially you could do.03:13
puffin444This is what I did.03:13
trailblazerthat seems a good idea03:13
puffin444What I don't know is when to do the actual pull request. I have some questions about how my code fits into the API03:13
puffin444I don't know if I should submit the pull request right away and ask then, or ask on IRC/mailing lists and then submit the PR after talking with them03:14
trailblazeri see , i guess mentors can better answer that question03:14
puffin444I just don't want to bother them with half-baked additions03:14
trailblazerso what are you studying?03:15
puffin444CS right now, probably at Robotics in the fall03:15
puffin444how about you?03:16
trailblazersame here bachelors in CS03:16
trailblazeri did ML course by stanford thats what got me hooked up in this area03:17
puffin444That's awesome!03:17
puffin444Well goodluck on your application!03:26
-!- puffin444 [230bf329@gateway/web/freenode/ip.35.11.243.41] has left #shogun []03:27
-!- pluskid [~chatzilla@60.191.2.230] has joined #shogun05:10
-!- pluskid [~chatzilla@60.191.2.230] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120312181643]]05:30
-!- trailblazer [18e9e06a@gateway/web/freenode/ip.24.233.224.106] has quit [Quit: Page closed]06:00
-!- vikram360 [~vikram360@117.192.175.5] has joined #shogun06:51
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun08:03
-!- hpsoar_ [7c108901@gateway/web/freenode/ip.124.16.137.1] has joined #shogun08:37
-!- PSmitAalto [82e9b263@gateway/web/freenode/ip.130.233.178.99] has joined #shogun08:57
n4nd0sonney2k: hi! are you around?09:06
n4nd0sonney2k: I have a doubt ... it turns out that there is an object, CLabels* created in the method classify_one_vs_one09:07
n4nd0sonney2k: the method returns that object, so it is not responsible of freeing it09:08
n4nd0sonney2k: however this method also does SG_REF(labels)09:08
n4nd0sonney2k: and if I use that method from another file, I am trying right now with an example, and also do SG_REF(labels)09:09
n4nd0sonney2k: then there is a memory leak09:09
n4nd0sonney2k: if I remove the SG_REF from the example then it disappear09:09
n4nd0sonney2k: but it seems to me more reasonable that it is the library the one that shouldn't do SG_REF because it is not responsible of freeing it, what do you think?09:10
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun09:25
n4nd0blackburn: hey!09:29
blackburnn4nd0: could you please describe this problem a little more?09:29
blackburnhey09:29
n4nd0blackburn: I was going to ask you about it :)09:30
blackburnif I got you right09:31
blackburnit should work in the following way09:31
blackburnany method returning new object should REF it before09:31
blackburnthen you don't need to REF this object cause it has 1 ref already09:31
blackburnall you need to destroy it is to UNREF it09:31
blackburnthen ref counter becomes 0 it is automagically destroyed09:32
n4nd0ok09:33
n4nd0then the thing I was doing wrong is to ref it after09:33
n4nd0do you know why is it correct to do it like this and not the other way?09:34
blackburnhmm yes09:34
n4nd0the other way = ref and unref where the method is callexd09:34
blackburnwell we should rather ref all the objects automatic09:35
blackburnand unref manually09:35
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]09:35
blackburne.g. how can you SG_REF in python?09:35
n4nd0aham09:36
n4nd0I see it09:36
n4nd0blackburn: ok thank you very much man :)09:36
blackburnyou are welcome09:36
blackburnoh I got late for 1.5 hours09:37
n4nd0blackburn: btw, it turned out not to solve the seg fault09:37
n4nd0blackburn: ups ... I can tell you later then09:37
blackburnno problem09:37
blackburnn4nd0: what turned out not to solve?09:37
n4nd0blackburn: the init_machines_for_apply at the end of train_machine09:38
blackburnah09:38
n4nd0blackburn: I have not investigated why, but it didn't solve it09:38
n4nd0blackburn: so I am still doing init_machines_for_apply(NULL) at the very beginning of aplly()09:38
blackburntry to check what in NULL09:38
n4nd0apply(), sorry09:38
n4nd0the features of each of the trained machines09:39
blackburnhmm does this all the job?09:39
blackburnno segfault?09:39
n4nd0of m_machine09:39
n4nd0yes, like that, it works just fine09:39
blackburnok then09:39
n4nd0another thing here09:39
n4nd0the constructor without argument of LinearMulticlassMachine09:39
n4nd0I think it brings problems09:40
n4nd0because if one uses it, then m_machine cannot be set09:40
n4nd0and it screws quite a bit ...09:40
blackburnuhh.09:40
blackburnwell actually it shoudln't be used09:40
n4nd0train can be called because the object m_machine is created with the default constructor of CMachine09:41
blackburnbut should safe though09:41
n4nd0but it turns out to seg fault that way09:41
blackburnn4nd0: just block all the things in this case09:41
n4nd0blackburn: block, with an SG_ERROR??09:41
blackburnyes09:41
blackburnI mean if someone would like to train machine without machine09:42
n4nd0yeah09:42
blackburnmay be is_ready() stuff should be modified09:42
n4nd0I tried to do that09:42
n4nd0the easy thing to do if (!m_machine)09:42
n4nd0does not work09:42
n4nd0because there is m_machine, the default constructor initializes it09:42
blackburnwith value??09:43
n4nd0I don't understand that sorry :S09:43
n4nd0it initializes it like CMachine()09:43
n4nd0with the default constructor09:43
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun09:43
blackburnthere is no initialization of m_machine actually09:44
blackburnshogun / src / shogun / machine / MulticlassMachine.cpp : 2009:44
blackburnshould be therre09:44
blackburnm_machines as well btw09:44
n4nd0oh fuck that is right09:44
n4nd0I made a mistake because CMachine is called09:45
n4nd0but it is called because MulticlassMachine is also a machine ...09:45
blackburnI guess that is the problem09:45
blackburnthen you can check if m_machine is here09:45
n4nd0so09:45
n4nd0in the default constructor09:45
n4nd0if I just set m_machine to null09:45
n4nd0it would do the work right?09:45
blackburnyes should09:46
n4nd0great :)09:46
blackburnif I do not get crazy09:46
n4nd0let's hope not ;)09:47
n4nd0another issue09:49
n4nd0MulticlassMachine.cpp:4009:49
n4nd0well wait, I probably not talking of the same line number as you can see09:49
blackburnclear_machines()?09:49
n4nd0but SG_UNREF(m_machine) in the destructor09:50
blackburnyes09:50
n4nd0valgrind detects that as invalid read of size 409:50
n4nd0I guess that this must be because this is already unreferenced in another place?09:50
blackburnwhat is the case?09:51
wikingblackburn: hey sorry but did you get the point of alex's email?/09:51
blackburnwiking: oh I glanced over it and postponed to read thoroughly a little bit later09:51
blackburnso not really :D09:52
wikingehhehe ok09:52
wikinglet me know since i really don't get some stufff09:52
n4nd0blackburn: in an example, create a machine*, create a multiclassMachine* with using the previous one09:52
blackburnI'll just answer in this thread09:52
blackburnn4nd0: oh and it fails??09:52
n4nd0blackburn: it doesn't really fail09:53
-!- PSmitAalto [82e9b263@gateway/web/freenode/ip.130.233.178.99] has quit [Ping timeout: 245 seconds]09:53
blackburnthat's a bit strange09:53
n4nd0blackburn: it is just that valgrind complains that there is an error when doing SG_UNREF in ~CMulticlassMachine()09:53
n4nd0maybe I am doing UNREF in my example and I shouldn't09:54
blackburnUNREF of?09:54
n4nd0I am actually doing UNREF in the example, yes, of the machine*09:54
blackburnhmm then you destroy machine09:54
blackburnit should be unrefed from the top09:55
n4nd0http://snipt.org/uggif109:55
n4nd0that is a part of the example09:55
blackburnaha09:56
blackburnlooks legal09:56
n4nd0mmm then09:57
n4nd0the unref shouldn't be in the destructor of CMulticlassMachine??09:57
blackburnshould be I think09:57
n4nd0then there must another something wrong mmmm09:58
blackburnyou didn't ref svm actually09:58
blackburnbut refed mc_svm09:58
n4nd0let me check it again, some other things changed so maybe it was a problem somewhere else09:58
blackburnI don't know whether it caused the problem09:58
n4nd0ahm09:58
n4nd0let me see09:59
blackburnbut I am 90% sure m_machine should be unrefed09:59
blackburnin mc machine09:59
n4nd0ok10:00
n4nd0you can be 100% :)10:00
n4nd0it was the missing SG_REF in the example probably10:00
blackburnnice10:01
n4nd0I think I am almost done with the job for one-vs-one then10:02
blackburnthat's nice10:02
blackburnactually I came up with reproducing case10:02
n4nd0what do you mean?10:02
blackburnimagine you have a machine (binary one)10:02
blackburnand create some multiclass machine based on that in some function or so10:03
blackburnonce it is done, multiclass machine is destroyed10:03
blackburnbut machine REF counter is still has extra 110:03
blackburnremove *is*10:03
blackburn:D10:03
blackburnso machine won't be destroyed at all!10:04
n4nd0mmm I see10:04
n4nd0but10:06
n4nd0I don't know if you mean that this is a potential error10:06
n4nd0or?10:06
blackburnyes10:07
blackburnso all dependencies should be free'd once they are not needed any more10:07
-!- mayanks43 [~marcus@115.248.130.148] has joined #shogun10:07
n4nd0but if the machine already existed, before creating the multiclass machine10:07
n4nd0it might be that it is needed somewhere else10:07
blackburnsure10:08
blackburnSG_UNREF will not delete it then10:08
n4nd0but that is good10:09
n4nd0is not a problem10:09
blackburnyes but w/o SG_UNREF10:09
blackburnmachine would have 2 references10:09
blackburneven when multiclass machine is gone10:09
blackburnand it won't be deleted10:09
n4nd0ah ok10:09
n4nd0you are just ensuring that it is good to do the SG_UNREF right?10:09
blackburnyes10:10
blackburn:D10:10
n4nd0ok :)10:10
n4nd0my bad10:10
n4nd0I was thinking that it was a problem that arose *leaving* the SG_UNREF in ~CMulticlassMachine10:10
blackburnnono it should be here10:11
n4nd0cool10:11
n4nd0btw, I am just curious10:13
n4nd0not so long time ago I heard about smart pointers10:13
blackburnyeah10:14
n4nd0woulnd't that make things easier?10:14
n4nd0I don't know that much about them thoug! :S10:14
blackburnI do not know about swig and these pointers..10:15
n4nd0aham, I understand10:15
-!- mayanks43 [~marcus@115.248.130.148] has left #shogun []10:15
n4nd0it seems that it is something new in C++10:16
blackburnit is ok to be gay^W^W^W to use smart pointers but we should always keep swig in mind10:16
n4nd0hahaha10:16
n4nd0I was just curious ... c'mon don't call be gay for that :P10:16
n4nd0call me*10:17
blackburnI just recalled a good old song lol10:17
blackburnbtw sonney2k rejected me when I asked him to marry me10:17
blackburneven this song didn't help10:17
n4nd0wtf?!10:18
n4nd0don't you have a gf? :P10:18
n4nd0which song is that?10:18
blackburnso I decided to go to siberia and start dating with bear10:18
blackburnhttp://www.youtube.com/watch?feature=player_detailpage&v=3j4t185wl-010:20
n4nd0:O10:23
blackburnoh sh~ I have to get to work10:23
blackburnsee you alter10:23
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed]10:23
n4nd0ok10:26
-!- harshit_ [~harshit@59.177.197.85] has joined #shogun10:31
harshit_hey n4nd010:32
n4nd0hey harshit_, qu? tal? ;)10:32
harshit_bian,y tu?10:32
n4nd0bien tambi?n10:33
harshit_hey i was thinking to make a new example in matlab : by using 20NewsGroup dataset on linearSVM10:33
harshit_what do you think should i do it ?10:33
n4nd0harshit_: yeah sure10:34
harshit_i thought of doing it in libshogun but i dont know how to open *.mat files in C++10:35
harshit_do you hav any idea on how to do that, coz 20Newsgroup datase is available in matlab file only10:36
n4nd0I am going to push an example doing that10:36
harshit_oh gr810:36
n4nd0but the issue is that the file I am using has no format10:36
harshit_that would be useful10:36
harshit_ohh10:36
n4nd0is just the vector values directly10:36
n4nd0the ones you are going to use has format?10:36
n4nd0I am downloading the data set to take it a look10:37
harshit_yes, matlab stores its matrix and objects in some specific format10:37
harshit_i dont think your example could be used on *.mat files10:37
n4nd0yeah ... then we cannot do that :(10:38
n4nd0but it is something we shoould learn probably10:38
n4nd0we have to ask blackburn or sonney2k about that10:38
n4nd0eh but wait10:38
harshit_yeah.. i'll try to figure that out and if possible i can get an pull out of it10:39
harshit_what10:39
n4nd0the don't actually have any specific format10:39
n4nd0I am reading the files .data .label with a text editor10:40
n4nd0and no problem10:40
n4nd0so I think you can do that with CStreamingSparseFeatures10:41
n4nd0because the representaion of the training data is sparse10:41
n4nd0and the labels with CStreamingSimpleFeatures10:41
n4nd0in the example I am going to push soon I use CStreamingSimpleFeatures10:42
n4nd0so it should be ok to follow as a reference10:42
n4nd0in any case ...10:42
n4nd0it turned out to be kind of painful for me to do it ... I am not completely sure I got it completely right done10:42
harshit_yeah you are right for .label and .data files10:43
harshit_but10:43
n4nd0so I suggest you to do it first in matlab/octave or python (you can read .mat with python very easy too)10:43
harshit_there are also other *.mat files need some way to be used in C++ .no ?10:43
n4nd0yes, I totally agree10:44
harshit_yeah exactly, thats what i thought, first in matlab and then in C++10:44
n4nd0don't know the general way to do it with shogun (don't really know if there is one already actually)10:44
harshit_okay so lets do some work ..10:44
n4nd0:)10:44
harshit_hey one last thing .. i have read a lot of differences between static and modular interfaces but actually till now i dont hav a clear idea of what is the main difference between them10:47
n4nd0I don't know exactly either10:54
n4nd0I know that static has limitations10:54
n4nd0for example it can just one classifier for each program (I think blackburn told me this)10:55
harshit_oh thats actually confusing11:07
-!- harshit_ [~harshit@59.177.197.85] has quit [Ping timeout: 264 seconds]11:19
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has quit [Quit: Leaving]11:25
-!- flxb [~cronor@e178172254.adsl.alicedsl.de] has joined #shogun12:00
-!- harshit_ [~harshit@59.177.197.85] has joined #shogun12:07
-!- vikram360 [~vikram360@117.192.175.5] has quit [Ping timeout: 265 seconds]12:23
-!- hpsoar_ [7c108901@gateway/web/freenode/ip.124.16.137.1] has quit [Quit: Page closed]12:37
-!- flxb [~cronor@e178172254.adsl.alicedsl.de] has quit [Quit: flxb]12:43
-!- harshit_ [~harshit@59.177.197.85] has quit [Ping timeout: 240 seconds]12:46
wikingasdf12:48
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has joined #shogun12:55
-!- vikram360 [~vikram360@117.192.187.98] has joined #shogun13:54
-!- flxb [~cronor@fb.ml.tu-berlin.de] has joined #shogun14:01
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]16:11
-!- flxb_ [~cronor@141.23.80.206] has joined #shogun16:51
-!- flxb_ [~cronor@141.23.80.206] has quit [Remote host closed the connection]16:52
-!- flxb_ [~cronor@fb.ml.tu-berlin.de] has joined #shogun16:52
-!- flxb [~cronor@fb.ml.tu-berlin.de] has quit [Read error: Connection reset by peer]16:53
-!- flxb_ is now known as flxb16:53
-!- harshit_ [~harshit@182.68.160.94] has joined #shogun17:47
-!- flxb [~cronor@fb.ml.tu-berlin.de] has quit [Quit: flxb]18:09
-!- harshit_ [~harshit@182.68.160.94] has quit [Ping timeout: 260 seconds]18:44
-!- blackburn [~qdrgsm@109.226.113.113] has joined #shogun18:46
-!- flxb [~cronor@e178169201.adsl.alicedsl.de] has joined #shogun19:01
-!- vikram360 [~vikram360@117.192.187.98] has quit [Read error: Connection reset by peer]19:22
-!- vikram360 [~vikram360@117.192.187.98] has joined #shogun19:23
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun19:24
blackburnn4nd0: main difference between static and modular19:25
blackburnis that later rocks19:25
blackburnwhile first one sucks19:25
blackburn:D19:25
n4nd0haha19:25
blackburnlatter*19:25
n4nd0I knew you were going to say that static sucks :P19:25
blackburnthe only reason (for me) to keep static is matlab suppor19:26
n4nd0but what are the reasons why it does suck?19:26
blackburnhave you seen static scripts?19:26
blackburnpretty poor capabilities19:26
blackburnonce you add new parameter you HAVE to implement this in static19:27
n4nd0I don't see the last point19:27
blackburnn4nd0: check shogun/gui folder19:31
n4nd0blackburn: I don't see that directory man19:34
blackburnui?19:34
blackburn:)19:34
n4nd0??19:35
blackburnshogun/ui?19:35
blackburnis there ui?19:35
blackburn:)19:35
n4nd0mm don't see that19:36
blackburnsrc/shogun/ui19:37
n4nd0got it :)19:39
blackburnn4nd0: so each time you want to add new feature19:45
blackburnyou HAVE to implement it in shogun/ui19:45
n4nd0aha19:45
n4nd0you mean for static right?19:45
blackburnyes19:45
blackburnwhile modular is a bleeding edge stuff hah19:45
n4nd0I understand19:46
-!- PhilTillet [~Philippe@38.124.160.27] has joined #shogun20:01
PhilTilletHey :p20:01
PhilTilletsonney2k, surprise in your mailbox!20:01
blackburnsonney2k: here?20:07
-!- PhilTillet [~Philippe@38.124.160.27] has quit [Ping timeout: 248 seconds]20:07
-!- PhilTillet [~Philippe@38.124.160.27] has joined #shogun20:12
-!- PhilTillet [~Philippe@38.124.160.27] has quit [Read error: Connection reset by peer]20:14
CIA-64shogun: Sergey Lisitsyn master * r777e558 / (2 files): Some final fixes for #401 (+11 more commits...) - http://git.io/dENg2w20:20
-!- vikram360 [~vikram360@117.192.187.98] has quit [Ping timeout: 244 seconds]20:24
-!- vikram360 [~vikram360@117.192.161.171] has joined #shogun20:24
shogun-buildbotbuild #636 of libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/libshogun/builds/636  blamelist: fernando.iglesiasg@gmail.com20:25
-!- blackburn [~qdrgsm@109.226.113.113] has quit [Ping timeout: 246 seconds]20:25
-!- blackburn1 [~qdrgsm@62.106.114.183] has joined #shogun20:26
shogun-buildbotbuild #637 of libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/libshogun/builds/63720:28
-!- blackburn1 is now known as blackburn20:31
shogun-buildbotbuild #576 of r_static is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/576  blamelist: fernando.iglesiasg@gmail.com20:31
blackburnohhh20:32
n4nd0this guy, shogun-buildbot, is always blaming me :(20:32
blackburnn4nd0: did not you implement it?20:32
n4nd0blackburn: they're with SG_NOTIMPLEMENTED20:33
blackburnthat's strange20:33
n4nd0I wanted to discuss with you again first :S20:33
blackburnI mean it should work20:33
blackburnhah no worries, I'll implement it20:34
blackburnah old build20:34
blackburnlets wait20:34
n4nd0old build?20:35
n4nd0ah ok, the number20:35
n4nd0maybe I should change my way of making commits ...20:36
n4nd0blackburn: should I do sth about it?20:42
n4nd0shogun-buildbot: work!20:48
shogun-buildbotWhat you say!20:48
shogun-buildbotbuild #577 of r_static is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/57720:50
n4nd0shogun-buildbot: that is nicer!20:51
shogun-buildbotWhat you say!20:51
blackburnn4nd0: yeah, all is ok20:54
n4nd0blackburn: did you need to do those methods yourself?20:54
blackburnn4nd0: kernel stuff?20:55
n4nd0blackburn: yeah20:55
blackburnno, feel free to do that if you want :)20:55
n4nd0blackburn: ok ;)20:55
blackburnbut it is not a lot of work20:55
n4nd0not at all20:55
blackburnit is up to you20:55
blackburnso do you want to do that?20:55
n4nd0yeah sure20:55
blackburnall you need is to implement set_subset methods for kernel20:55
blackburnfor custom it should set_row_subset and set_col_subset20:56
blackburnfor simple kernels it should add subset to features20:56
blackburnsame with remove_subset20:56
n4nd0for simple is like for the LinearMachine20:56
blackburnyes20:57
n4nd0it will be ready soon with some example of use ;)20:58
blackburnok thanks20:59
shogun-buildbotbuild #406 of java_modular is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/java_modular/builds/406  blamelist: fernando.iglesiasg@gmail.com21:09
CIA-64shogun: Sergey Lisitsyn master * rafb8b5e / (src/CONTRIBUTIONS src/NEWS): Updates for NEWS and CONTRIBUTIONS - http://git.io/I9FuYg21:19
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 265 seconds]21:29
-!- wiking_ [~wiking@huwico/staff/wiking] has joined #shogun21:44
-!- wiking [~wiking@huwico/staff/wiking] has quit [Read error: No route to host]21:44
-!- wiking_ is now known as wiking21:44
-!- wiking [~wiking@huwico/staff/wiking] has quit [Client Quit]21:47
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun21:49
gsomixhi21:53
gsomixblackburn, yo.21:53
blackburnhi21:53
shogun-buildbotbuild #402 of python_modular is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/402  blamelist: fernando.iglesiasg@gmail.com21:55
blackburnnaah21:56
n4nd0:(21:59
n4nd0./classifier_averaged_perceptron_modular.py??22:00
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun22:03
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 260 seconds]22:18
shogun-buildbotbuild #407 of java_modular is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/java_modular/builds/40722:40
n4nd0:)22:40
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun22:53
-!- gsomix [~gsomix@188.168.2.195] has quit [Ping timeout: 252 seconds]22:57
-!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Remote host closed the connection]23:04
shogun-buildbotbuild #403 of python_modular is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/40323:11
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has quit [Quit: Leaving]23:12
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]23:33
* sonney2k wakes up23:34
blackburnsonney2k: how do you do?23:35
blackburnsonney2k: we just talked to n4ndo, and I wasn't able to recall why do you hate STL?23:36
@sonney2kI feel every muscle / bone today :)23:37
blackburnsonney2k: trained yesterday?23:37
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]23:40
@sonney2kblackburn, went to the zoo with the 'little' one23:42
blackburnah23:42
@sonney2kcarrying him all day23:42
@sonney2k+ backpack23:42
blackburnI see23:42
@sonney2kwhat does n4ndo need STL for?23:43
blackburnsonney2k: we didn't come up with any example yet :D23:44
blackburnbut actually there could be places where it would fit better23:44
@sonney2k10 proposals...23:44
blackburnonly 2 more today?23:44
blackburnsonney2k: do you think it is ok to provide Kernel LLE capabilities in LocallyLinearEmbedding class?23:45
blackburnI mean to remove such duplication23:45
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun23:45
blackburnnow we have both23:46
@sonney2kI would keep both...23:46
blackburnsonney2k: hmm ok23:46
-!- flxb_ [~cronor@g231240112.adsl.alicedsl.de] has joined #shogun23:52
-!- flxb [~cronor@e178169201.adsl.alicedsl.de] has quit [Ping timeout: 246 seconds]23:55
-!- flxb_ is now known as flxb23:55
blackburnsonney2k: so is there any actual reason to avoid STL?23:56
@sonney2kexcept that I don't like it you mean?23:56
@sonney2kprobably not23:57
blackburnok23:57
@sonney2kblackburn, what was that SG_REF / UNREF issue you had with n4nd0?23:58
blackburnsonney2k: he forgot to SG_REF svm that he was using23:58
blackburnlater this svm was passed to generic multiclass thing23:59
@sonney2kyeah usually that is the issue or that something is not inited with NULL23:59
blackburnall is ok now23:59
@sonney2kor forgotten SG_REF when returning stgh23:59
@sonney2ksth23:59
--- Log closed Fri Mar 30 00:00:19 2012