Open in new window / Try shogun cloud
--- Log opened Sat Oct 27 00:00:17 2012
@sonney2kblackburn, look may 7 2012 seems to have been the most active at #shogun00:10
@sonney2kis that due to us enabling the bots maybe?00:11
blackburnsonney2k: gott! we were talking all day long00:12
@sonney2kyeah crazy everyone was active that day00:12
@sonney2kblackburn, I just got a confirmation from Carol - our application got submitted00:13
blackburnsonney2k: nice00:13
blackburnsonney2k: I am in MLdepression :(00:14
blackburnwe have to extend shogun somehow00:14
n4nd0Bayesian learning!00:16
blackburnn4nd0: I am too stupid to understand bayesian stuff00:17
n4nd0c'mon you are not00:18
n4nd0but seriously, I think Bayesian stuff can be really cool00:18
blackburnno idea00:18
n4nd0although I must admit I have just seen a few fundamentals of it so far00:18
blackburnI fucked up my brain with alexandrescu00:19
blackburncan't see shogun code anymore00:19
n4nd0what do you want to do about it?00:19
blackburnmay be lay somewhere and cry, may be travel around siberia00:20
blackburnnot sure what can i do with it00:20
@sonney2kblackburn, write some nice example!00:24
@sonney2kblackburn, I have something concrete in mind00:24
blackburnsonney2k: they are all useless :(00:24
blackburnwhat is your idea?00:24
@sonney2ksomthing that is live on the website basically00:24
@sonney2kso just some 2-class svm but writting in java script / python django00:25
@sonney2k+ utilizing shogun underneath00:25
@sonney2kand some fancy visualization via d3js00:25
blackburnis it possible to call pythno from java script?00:26
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.] has quit [Quit: Page closed]00:28
@sonney2kblackburn, sure on our website w/ dango00:30
@sonney2kI think such kind of demos would make it much easier for people to grasp what shogun is about00:31
@sonney2kand we could have a toy demo for each class of shogun's algorithms00:31
@sonney2kblackburn, for other serious demos we really need concrete examples. I think the gender detector based on opencv would also be cool00:32
@sonney2kor some of wiking's live image segmentation00:32
blackburnsonney2k: there is no machine learning in gender detection nor segmentation nor anything machine vision00:33
blackburnI mean it is just learn a model and apply00:33
blackburnsonney2k: in opencv they even include "w" vector for pedestrian detection00:34
blackburnno machine learning anymore at all00:34
blackburnsonney2k: are your kids better?00:35
@sonney2kblackburn, huh? What do you mean? The w vector needs to be learned so why is it not ML?00:36
@sonney2ksleeping now00:36
blackburnsonney2k: one don't need a huge library to learn just w00:36
blackburnand in the end nothing matters in ml sense, it is rather what you do with features00:37
blackburnyeah I am in marvin state again00:37
@sonney2kso marvin tell me which field do you want to kill next?00:41
@sonney2kand actually sure you do - for the gender detection you need coffin or you won't be able to learn00:42
blackburnsonney2k: liblinear doesn't use coffin00:45
@sonney2kblackburn, it does00:46
blackburnin shogun?00:46
blackburnbut in liblinear?00:46
@sonney2kbtw we totally didn't mention in NEWS that the gsoc's student's projects are in00:46
@sonney2kin liblinear nope - that is the problem00:47
blackburnyeah we totally didn't mention 50% of things we should better00:47
@sonney2kblackburn, what else?00:47
blackburnsonney2k: but it is not a problem for other people - that is the problem I am depressed with00:47
blackburnsonney2k: SLEP fuckup at least00:47
blackburnto remember :)00:47
@sonney2kwe forgot heikos stuff too00:47
@sonney2kblackburn, sure it is - only very few people have 512G mem and hundreds of cores00:48
blackburnsonney2k: btw about coffin00:48
blackburnI think it could be implemented using generic programming00:48
@sonney2kand with shogun and that amount of mem they could have used orders of magnitude more examples and get much higher accuracy00:48
blackburndense and add are functors00:48
blackburndidn't try but conceptually it could simplify code00:49
@sonney2kno idea about that00:49
@sonney2kblackburn, I think you should focus on submitting your edrt to jmlr00:49
blackburnI am working on it00:50
blackburnjust now00:50
@sonney2kyou have too many things going on00:50
blackburnsonney2k: btw n4nd0 will be in (I hope so)00:50
@sonney2kfor what?00:50
blackburnsonney2k: edrt00:50
@sonney2kwhat did he do?00:50
blackburnsonney2k: nothing yet - he will help me with porting to eigen3 and new algorithms00:51
@sonney2khmmhh, don't make it too big otherwise you will never finish00:51
@sonney2kI recall that reviewers were happy already last time00:52
@sonney2kblackburn, look at this
@sonney2kand then click group00:53
@sonney2kisn't that cool00:53
@sonney2kthat is d3js :)00:53
blackburnyeah it is00:53
@sonney2kblackburn, or
@sonney2kdrag around some node00:53
@sonney2kso that is what I want for shogun demo00:54
@sonney2kblackburn, that is the mode you are in right
blackburnom nom nom01:18
blackburnsonney2k: exactly01:18
@sonney2kblackburn, hmmhh well your edrt spin off - well do whatever you want01:48
blackburnsonney2k: ?01:49
blackburnsonney2k: I want to develop it standalone and then bind shogun to newest version01:49
blackburnsonney2k: what did you mean?01:52
@sonney2kI know.01:54
@sonney2kexactly that01:54
@sonney2kI still dont' get why but hey do whatever you want01:54
blackburnsonney2k: do you see any other way for that paper?01:54
blackburnI can't demarkate where shogun ends01:54
@sonney2ksure. they didn't care about that01:55
@sonney2kall you have to do is explain just edrt's features and not so much shogun's01:55
blackburnI think cheng meant that01:55
@sonney2kno cheng complained about that you described shogun's features01:56
blackburnsonney2k: it is just like libqp01:57
@sonney2kblackburn, I don't understand your gsoc summary02:13
@sonney2k" During the summer Sergey has ported a few algorithms from the SLEP and02:14
@sonney2kMALSAR libraries with further extensions and improvements.02:14
blackburnehmm what is wrong02:14
@sonney2kdidn't you implement some multitask stuff your own?02:14
blackburnsonney2k: from matlab code mostly02:15
@sonney2kblackburn, can you give me the links to slep/malsar please?02:15
@sonney2kblackburn, from chris or whom?02:15
blackburnsonney2k: SLEP02:15
blackburnand MALSAR02:15
blackburnsonney2k: they have a lot of matlab code - I adopted their C code and developed C++ code using the matlab code02:15
blackburnbut with coffin and all the stuff02:16
blackburnlatest are using eigen3 as well with big impact on performance02:16
@sonney2kcurrently your summary doesn't sound impressive02:16
@sonney2kthough what you say here is cool so please tweak.02:17
@sonney2kname the algorithms you ported explicitly and what you did to them02:17
blackburnwell I didn't do anything special this summer02:17
@sonney2kI think that will make it sound better and gets closer to the truth02:17
blackburnpretty bad work actually02:18
@sonney2kas usual02:18
@sonney2kblackburn, name them please02:18
blackburnsonney2k: there will be too many algorithms named02:18
@sonney2kthat is good02:19
@sonney2kand needed02:19
@sonney2kand please mention matlab -> c++ / coffin eigen3 too02:19
blackburnSergey Lisitsyn was our GSoC student for the second time and he focused on multitask learning. Multitask learning is a modern approach to machine learning that learns a problem together with other related problems at the same time using a shared representation. This approach often leads to a better model for the main task, because it allows the learner to use the commonality among the tasks. During the summer Sergey has developed efficient implementatio02:24
blackburnsonney2k: ^ ?02:24
blackburnI'd actually prefer 'He's spent some time on useless things nobody will ever touch'02:24
@sonney2kblackburn, your project description is currently worst of all :/02:28
@sonney2kand that is really is because of your stupid marvin mood02:29
blackburnno idea what to do with it02:29
@sonney2kblackburn, *please give me the list of algoritms you ported*02:29
blackburnit is already in text I  sent02:29
blackburnNamely, they are L12 group tree and L1q group multitask logistic regression and least squares regression, trace-norm multitask logistic regression, clustured multitask logistic regression, basic group and group tree lasso logistic regression.02:30
blackburnsonney2k: ^?02:30
@sonney2kwhen did you sent that where?02:30
blackburnirc filtered it out probably02:30
blackburnSergey Lisitsyn was our GSoC student for the second time and he focused on multitask learning. Multitask learning is a modern approach to machine learning that learns a problem together with other related problems at the same time using a shared representation. This approach often leads to a better model for the main task, because it allows the learner to use the commonality among the tasks.02:31
blackburnDuring the summer Sergey has developed efficient implementations of various multitaks algorithms using the SLEP and MALSAR libraries. Namely, they are L12 group tree and L1q group multitask logistic regression and least squares regression, trace-norm multitask logistic regression, clustured multitask logistic regression, basic group and group tree lasso logistic regression. All the implemented algorithms use COFFIN framework for flexible and efficient l02:31
blackburnsonney2k: works this time?02:31
@sonney2kyes now I received it02:31
@sonney2kblackburn, but cut off02:32
@sonney2kwhat is following efficient I02:32
blackburn2nd message?02:32
@sonney2kyes the "All the implemented..." is not complete02:32
@sonney2ksentence I mean02:33
blackburnAll the implemented algorithms use COFFIN framework for flexible and efficient learning and some of the algorithms were implemented using efficient Eigen3 library.02:33
@sonney2kthat's it then right?02:33
blackburnstupid irc02:33
@sonney2know it sounds much better02:33
@sonney2kblackburn, what is clustured multitask ?02:34
blackburnsonney2k: tasks are clustered02:34
@sonney2kor structured02:34
blackburnsonney2k: you just say you have 81 tasks and 9 clusters02:35
blackburnand it finds you the clusters02:35
blackburnand solution as usual02:35
blackburnrelaxed k-means02:35
blackburnstupid idiot thing no idea why ever use it02:35
@sonney2kyes marvin02:37
* sonney2k hands blackburn a cookie02:41
blackburncookies are useless02:42
blackburnmy gf got bored and wrote a QNX problem, why on the earth one should write programs for QNX02:42
@sonney2kblackburn, what other big changes did we have in shogun02:44
@sonney2kI recall SGVector etc + refcounting02:44
@sonney2kand utilizing eigen302:44
blackburnsonney2k: hmm02:45
blackburnlet me recall02:45
@sonney2kblackburn, I am reading what we have achieved this summer and I cannot *not at all* see the slightest reason to be depressed02:48
@sonney2kto the contrary02:48
@sonney2klooks like a heroic effort02:48
blackburnsonney2k: yeah in number of lines probably02:49
blackburnsonney2k: a few useless refactoring operations were carried out as well02:51
blackburnwe now have even more folders02:51
@sonney2khaha :D02:52
blackburnokay time to sleep02:53
blackburnsonney2k: have a nice weekend and good night02:54
@sonney2kblackburn, sweet as in sugar dreams!02:54
-!- blackburn [~blackburn@] has quit [Quit: Leaving.]02:56
-!- ptizoom [] has quit [Quit: Ex-Chat]04:32
-!- ptizoom_ [] has joined #shogun04:32
-!- ptizoom_ [] has quit [Quit: Ex-Chat]08:22
-!- ptizoom__ [] has joined #shogun08:22
-!- Marty28 [~marty@] has joined #shogun08:25
-!- Marty28 [~marty@] has quit [Quit: Colloquy for iPad -]08:35
-!- gsomix [~gsomix@] has joined #shogun08:36
gsomixgood morning08:37
-!- ptizoom__ is now known as ptizoom09:17
wikinggsomix: here?10:50
wikingactually you might be the person i need now :D10:50
wikingsonney2k: you awake? :)11:07
shogun-buildbotbuild #147 of nightly_default is complete: Failure [failed test]  Build details are at
-!- n4nd0 [] has joined #shogun13:23
-!- blackburn [~blackburn@] has joined #shogun13:41
wikingblackburn: ping14:06
wikingok so i've did like almost everything14:07
wikingbut still cannot get python to be debugged14:08
wikingi get a segfault if i start using Director class14:08
wikingthat's all i know :)14:08
blackburndid you try that gdb attach or anything like thaT?14:08
blackburnwiking: did you try valgrind?14:09
wikingi cannot attach14:09
wikingsince it's very fast14:09
blackburnit works for such debugging as well14:09
wikingvalgrind mmm14:09
wikinglet's see14:09
blackburnvalgrind catches segfaults as well14:09
wikingtrying :)14:10
wikingdidn't work :)14:10
wikingonly get segfault14:11
wikingand i can dig out this from apple's crash reporter14:11
wikingThread 0 Crashed:: Dispatch queue:
wiking0   org.python.python             0x000000010a6dd6b5 _PyType_Lookup + 14914:11
wiking1   org.python.python             0x000000010a6c9f6a _PyObject_GenericGetAttrWithDict + 17014:11
wiking2   org.python.python             0x000000010a68db18 PyObject_CallMethodObjArgs + 16814:11
wiking3                 0x000000010b12d10d SwigDirector_DirectorData::get_name() const + 173 (modshogun_wrap.cxx:10315)14:12
wiking4   libshogun.13.dylib            0x000000010c9f56fc shogun::CSGObje14:12
blackburnoh getname14:12
blackburndid you declare get_name too?14:12
wikingi'll kill nerdtree NOW14:14
wiking                virtual const char* get_name() const { return "DirectorData"; }14:14
blackburnI think I had that before14:18
blackburnwiking: can you show your python code?14:20
wikingblackburn: if i use Data as base class for DNA14:22
wikingeverything is ok14:23
blackburnwiking: ok the problem is that somewhere in C++ code get_name is called14:25
blackburnand swig got crazy with it somehow14:25
blackburncan't remembed what did I do with it last time14:27
blackburnwiking: ok I am near14:29
-!- wiking_ [] has joined #shogun14:31
-!- wiking [~wiking@huwico/staff/wiking] has quit [Read error: Connection reset by peer]14:31
blackburnfirst, we have to be sure only one thread is used14:34
wiking_ok how?14:34
wiking_the same14:39
wiking_Thread 0 Crashed:: Dispatch queue:
wiking_0   org.python.python             0x00000001019a96b5 _PyType_Lookup + 14914:40
wiking_1   org.python.python             0x0000000101995f6a _PyObject_GenericGetAttrWithDict + 17014:40
wiking_2   org.python.python             0x0000000101959b18 PyObject_CallMethodObjArgs + 16814:40
wiking_3                 0x00000001023fc10d SwigDirector_DirectorData::get_name() const + 173 (modshogun_wrap.cxx:10315)14:40
wiking_4   libshogun.13.dylib            0x0000000103cc46fc shogun::CSGObject::unref() + 156 (SGObject.cpp:174)14:40
wiking_5                 0x0000000102c804f5 _wrap_delete_DirectorData + 197 (modshogun_wrap.cxx:263538)14:40
wiking_6                 0x0000000102608028 _wrap_delete_DirectorData_closure + 56 (modshogun_w14:40
wiking_ok this i'll pastebin14:40
-!- wiking [] has joined #shogun14:42
-!- wiking_ [] has quit [Read error: Connection reset by peer]14:42
blackburnwiking: see the problem?14:42
blackburnwe call get_name of object that was deleted14:43
wikingsorry i'm trying to put together an osx buildbot14:43
wikingok but why it get's deleted i really wonder14:43
wikingand why it's not happening14:44
wikingwhen i don't use the director class?14:45
blackburnwiking: we have to put some flag or call some function or so14:45
blackburnI recall that14:45
blackburnbut what to call/set I do not recall14:45
blackburntrying to find14:45
blackburnwiking: call data.__disown__ after creation14:46
wikingwhere? :)14:47
blackburnwiking: so you create MotifData14:47
blackburnsomewhere in code14:47
blackburnand after instantiation14:47
blackburncall __disown__() method14:47
wikingok cool14:47
wikingnow it's not crashing14:48
blackburnwiking: some reference counting issue..14:48
wikingok but i have the same shiftfuck now14:48
wikingthat i get back a CData from features.get_sample(i)14:49
wikingbut i would need DNA or MotifPos14:49
blackburnwiking: oh you finally found what is obtain_from_generic for14:50
wikingand it seems there's no way around this14:50
wikingyeah but what if i specify a class14:50
wikingin python14:50
wikingor in this case this is not possible14:50
wikingand i should just not even try it :D14:50
wikingbecause if this is the case14:51
wikingthe whole concept of having latent svm in python is fucekd :)14:51
blackburnno, why?14:51
wikingso here's the deal14:51
wikingi need to define 3 things for a latent svm14:51
wikinglatent model14:51
blackburnyou get Data but you need a DNA?14:51
wikingi need DNA14:51
wikingsince it has some fields that i need in LatentModel14:52
wikingbut obviously Data doesn't have it14:52
blackburnin similar fashion it would be DNA.obtain_from_generic(m_features.get_sample(idx))14:52
blackburndon't you like it?14:52
wikingblackburn: but how do i do the typecasting in python/ :)14:52
wikingi mean i'll have to write the c++ for DNA class then14:52
wikingi mean i'll have to define DNA class and implement in c++14:53
wikingthat's what you are saying now14:53
blackburnfucking python14:53
wikinghehe hyeah14:53
blackburnwhy not to have casting?14:53
blackburnthey denied it allthrough14:53
blackburnwiking: so what to do with it..14:55
wikingi guess there's no way this will work then14:57
gsomixwiking: here14:58
wikinggsomix: hehehe we've just realised that it's not going to work out with blackburn15:00
wikingok so i have to reimplement now this in c P)15:00
wikingi was hoping that i can use soon python interface15:00
gsomixwiking: ok. it's comfortably, that there is blackburn.15:01
shogun-buildbotbuild #596 of deb3 - modular_interfaces is complete: Failure [failed test libshogun]  Build details are at  blamelist: Sergey Lisitsyn <>15:02
-!- heiko [] has joined #shogun15:27
blackburnwiking: I've no idea what to do with it still15:43
-!- K0stIa [] has joined #shogun15:46
-!- K0stIa [] has left #shogun []15:46
wikingblackburn: afaik it's not possible and that's it :(16:35
blackburnwiking: we should ask somewhere16:36
blackburnmay be guido16:36
blackburnwhy the heck you do it with us, guido16:36
blackburnwiking: having STL party in code16:38
wikingblackburn: hehehe well the new sosvm has stl as well16:39
wikingi still haven't got around changing it16:39
blackburnwiking: IIRC sonney2k has nothing against STD in .cpp16:39
wikingwhere? :)16:39
blackburnwell I think so16:40
wikingtoday might be the day that i can give an osx buildbot16:41
wikingand then we'll see the bug in StreamingFeatures as well16:41
wikingso somebody should fix that :)16:41
blackburnwe have to solve your problem somehow16:41
wikingit's funny that it's only osx who's memory manager fails on that bug16:41
wikingblackburn: yeah would be nice to be able to lsvm in python16:42
blackburnwiking: I started to fix that bug but had to change 100+ lines and gave up16:42
wikingso you acted like this:
wikingand shouted fuckthis shit :)16:43
wikingi'll join now #swig16:49
wikingif there's one and ask around16:49
wikingtitle: "Semantic Web Interest Group 24x7 publicly logged hack-'n'-chat."16:50
wikingi guess i'm on the wrong place :)16:50
wikinghehehe nice16:50
wikingin the topic: "ften confused with the Interface Generator, which is on channel #swig"16:50
wikingnice there are exactly 2 ppl in the #swig channel16:53
blackburnwiking: consider mailing list then :)16:57
wikingis it now intenional that MSG_ERROR will not cause and exit(1)17:01
wikingi.e. SG_ERROR("asdf");17:01
-!- heiko [] has quit [Quit: Leaving.]17:02
blackburnwiking: no idea17:14
wikingcurrently it's not killing the app17:24
wikingso basically we have a some stupid warnings because of this17:24
wiking in the static analyzer17:24
-!- heiko [] has joined #shogun18:42
-!- n4nd0 [] has quit [Read error: Operation timed out]19:08
@sonney2kwiking, no it should kill the app if we are not on a python etc interface19:13
wikingmmm weird thing19:24
@sonney2kwiking, ?19:24
wikingas scan-build detects it as if it's not killed19:24
wikingsonney2k: do you maybe have any solution ideas how one can typecast back to a derived class in python?19:34
wikingif i defined the derived class in python19:34
blackburnsonney2k: yeah we got catch 22 situation19:49
-!- gsomix [~gsomix@] has quit [Quit: Leaving]20:22
@sonney2kwiking, not possible20:51
-!- sonney2k [] has left #shogun ["Ex-Chat"]20:51
wikingblackburn: here's our answer :)20:54
blackburnwiking: amazing20:54
-!- heiko [] has quit [Quit: Leaving.]20:57
-!- heiko [] has joined #shogun21:04
-!- heiko [] has quit [Quit: Leaving.]21:31
-!- sonney2k [] has joined #shogun21:43
-!- mode/#shogun [+o sonney2k] by ChanServ21:43
blackburnsonney2k: ok to add iterators to features?21:48
blackburnsonney2k: with all you hate to stl do you realize stl-like for or for_each will work for any possible features with no changes at all (including streaming)21:49
@sonney2kblackburn, I have no idea about iterators in C++21:50
blackburnsonney2k: I believe it is a good pattern actually21:51
@sonney2kspeed trade-off?21:52
blackburnsonney2k: it is compile-time abstraction21:52
blackburnsonney2k: just like eigen3 stuff it goes away on compilation21:53
blackburnsonney2k: iterator is a pointer but it is generic21:54
blackburnwiking: agree? :D22:03
@sonney2kblackburn, did you measure?22:03
wikingsonney2k: if u find me a good way to convert dmg to iso then we can have an osx buildbot22:03
blackburnsonney2k: measure what exactly?22:03
wikingsonney2k: i've googled and tried everything in the afternoon but couldn't get working any of it22:03
blackburnsonney2k: okay let me show you assembly code in 5 minutes :D22:03
wikingblackburn: -Os the code :)22:04
@sonney2kwiking, what did you not get working?22:04
wikingsonney2k: couldn't get a dmg converted to iso22:04
wikingsonney2k: and i need an iso for VB22:04
wikingvb = virtualbox22:05
@sonney2kwiking, doesn't dmg2img work?
wikingsonney2k: tried22:09
wikingsonney2k: no luck22:09
wikingsonney2k: i've tried doing dmg2img && mount as hfsplus22:09
wikingbut couldn't mount the img22:09
wikinghad some problems... couldn't find hfsplus superblock22:10
@sonney2kwiking, well worst case burn the dmg to a dvd and then copy / create an iso from it22:10
wikingsonney2k: heheh no bw for upping now an iso from my place22:11
shogun-buildbotbuild #597 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at
-!- heiko [] has joined #shogun22:42
@sonney2kheiko, around?23:00
heikosonney2k, yes! hi23:00
@sonney2kheiko, I'd like to have your blog on shogun planet23:00
@sonney2knot the gsoc blog23:00
heikothe shogun one?23:00
heikosure go ahead!23:00
@sonney2kyour last entry didn't appear on planet23:01
heikoyes, I started a new one23:01
heikoshogun related23:01
wikingblackburn: where's asm? :D23:01
heikosonney2k, where I want to write stuff I do for shogun23:01
@sonney2kheiko, well what is the url for all?23:01
blackburnwiking: it looked similar and I gave up :D23:01
heikoEhm, I dont have one I think, let me construct one23:01
blackburnwith -O3 it is nearly the same23:01
heikoIll send it to you23:01
@sonney2kwell shogun planet doesn't necessarily mean only shogun stuff23:02
@sonney2kanything authors (or related people) write is totally fine23:02
@sonney2kcan be about weather or whatever23:02
heikosonney2k, ok23:02
@sonney2kbut missing your last blog post on planet is stupid23:02
heikosonney2k, indeed23:03
heikosonney2k, can you just add the other blog url feed?23:03
heikosonney2k, since I am pretty lame in webprogramming and dont know how to create a feed of the featured articles from the main site (using joomla)23:04
heikosonney2k, this is the url23:04
heikothe feed is on the bottom23:04
blackburnmatrix experts anyone?23:05
@sonney2kmuch better23:07
@sonney2kheiko, how do you get the tex stuff translated into formulas?23:07
heikosonney2k, yes its annoying that the graphics dont appear23:07
heikoits a joomla plugin23:07
blackburnY' * (W^-1) * Y23:08
heikoI think they are loaded from an external page23:08
heikoby the plugin23:08
blackburncan somebody translate that into equation to avoid inverse?23:08
@sonney2kheiko, too bad that this is not preserved in the feed23:08
heikosonney2k, indeed, Ill check for a solution to generate graphics instead23:09
@sonney2kheiko, there is this
@sonney2kheiko, you basically just write latex code and it is translated via javascript23:11
wikingocas_time 3082.422861 s23:11
wikinglet's see how it is :>23:12
heikosonney2k, ok I will take a look23:12
@sonney2kheiko, let me just add it to the website23:14
heikosonney2k, how to do that?23:14
heikoI actually have no idea about this stuff ;)23:14
@sonney2kheiko, well the mathjax doc says just to include a .js script23:14
@sonney2kthat's all23:15
@sonney2kso then as soon as you write latex math code it will be translated23:15
wikingchrome doesnotlike mathml23:15
@sonney2kheiko, and this would work on your joomla etc too23:15
wikingblackburn: ping23:15
blackburnwiking: pong23:15
wikingblackburn: whatyaneed?23:16
blackburnwiking: B = Y' inv(W) Y23:16
blackburnwant to have W without inv23:16
blackburncosts too much for me23:16
wikingblackburn: btw check out this one
wikingblackburn: press 'f' for fullscreen23:16
wikingblackburn: and of course you can do all the fancy css3 transformations ;)23:16
blackburnwiking: your work?23:17
wikingblackburn: noup of course not :)23:17
wikingblackburn: just put together 2 jquery plugins23:17
blackburnI see23:18
wikingblackburn: wekk i guess u r looking for B23:18
blackburnwiking: yes I need to translate it to WB = ..23:19
blackburnhmm may be it is possible in two steps?23:19
wiking /*W23:19
blackburnwiking: what does that mean?23:19
@sonney2kwiking, I've seen someone use that + d3js and mathjax23:19
@sonney2kvery fancy talk that you can afterwards easily put on the web23:20
wikingsonney2k: yep23:21
@sonney2kheiko, that is how you can use it in joomla
wikingblackburn: does Y have any special property?23:23
blackburnwiking: orthogonal23:23
blackburnwait wait so23:24
blackburnB = Y' inv(W) Y23:24
blackburnY B = inv(W) Y23:24
blackburnwiking: looks correct for you?23:25
wikingy' = transpose of y or?23:25
blackburnleft multiply by Y?23:25
blackburnY' is transpose23:25
blackburnY B = Y Y' inv(W) Y23:25
blackburncan we do that?23:25
blackburnI haven't been working with matrices for a while :D lol23:26
wikingmatrix mult is does not commute23:26
blackburnyes it is not23:26
heikosonney2k, I now changed the latex to cache the images locally, so it should be fine, how can I tell shogun planet to update itself?23:26
wikingunless they are diagonal and are the same dim23:26
blackburnyou mean it will be23:26
blackburnY B = Y (Y' inv(W) Y)23:26
wikingBY' = Y' inv(w) Y Y'23:27
wikingyeah it's almost the same :)23:27
blackburnokay BY' = Y' inv(W)23:27
blackburnbut it is not a linear equation system yet23:27
@sonney2kheiko, well even better just use mathjax23:27
blackburnwiking: do you remember whether we can transpose both left and right side?23:28
heikosonney2k, then I have to install it, I really would like to avoid that if it works otherwise23:28
@sonney2kheiko, look I just did this
blackburnwe can23:28
heikosonney2k, how worked that?23:29
@sonney2kheiko, no only write this one line in your doc23:29
heikobut dont I have to install the software before?23:29
@sonney2kthen you can immediately have latex code in there23:29
wikingblackburn: but A*A' != I23:29
heikoreally, just this one liner?23:29
heikocool thing23:29
@sonney2kheiko, no just follow
heikosorry I got that wrong23:29
blackburnwiking: what is A?23:29
wikingblackburn: any matrix23:29
@sonney2klooks cool right23:29
blackburnwiking: A*A' = I for an orthonormal matrix23:29
@sonney2kheiko, for that to work I replaced {tex} with \[ and {/tex} with \] on all pages ...23:30
wikingah yeah u said that it's orthonormal23:30
@sonney2kwhen n4nd0 is around I will tell him too23:30
heikosonney2k, I think n4nd0 is on wordpress where things work automatically23:30
blackburnwiking: hmm when we have BY' = Y' inv(W), we transpose both parts YB' = inv(W)' Y23:31
@sonney2kheiko, well look at how awful it looks on his blog23:32
@sonney2kso he should convert to using mathjax too23:32
wikingwell then altogether u have YBY' = inv(w)23:32
heikosonney2k, youre right, this mathjax is an awesome tool23:32
wikingmathjax is a shitty replacement of mathml23:32
blackburnwiking: inv(W)' Y can be easily computed23:32
blackburndenote it by K then we have YB' = K23:32
@sonney2kheiko, now imagine this used with
wikingit's just helping chrome and other shitty browsers who doesn't support mathml23:32
@sonney2kimpress js23:32
@sonney2kheiko, to give talks23:33
wikingsonney2k: jmpress.js is better23:33
wikingsonney2k: imho23:33
wikingsonney2k: as it's jquery based so u really can do some fancy shit23:33
wikingi've started with impress.js23:33
wikingbut for some reason bartaz doesn't want jquery in it23:33
wikingso that's why there's jmpress.js23:33
heikosonney2k, that one is pretty cool!23:34
blackburnsonney2k: impressive, it is a next step after beamer ones23:34
blackburnso evolution is god damn shit powerpoint -> beamer -> js :D23:34
blackburnwiking: thanks I think it should work23:34
blackburnwiking: I am trying to leverage redsvd for my dirty purposes23:35
@sonney2kheiko, then combine this with d3js for figures
@sonney2kheiko, e.g.
@sonney2kheiko, or
wikingbattlestation down :(23:36
wikingsonney2k: !!!23:37
wikingsonney2k: irc hook for github23:37
wikingsonney2k: can u fire it up plz?23:37
@sonney2kwiking, cia is dead23:37
wikingsonney2k: but it's not based on cia23:37
@sonney2kand I tried to configure the standard irc hook for github but it doesn't work23:37
@sonney2kwiking, no idea how to get it to work23:38
wikingsonney2k: r u sure u used shogun-toolbox account and not yours?23:38
wikinglike in case of travis23:38
@sonney2kwiking, sure23:38
@sonney2kwiking, try to add notifications to your local fork23:38
@sonney2kif it works we get this to work23:39
wikingnotice - Enables notice support. Make sure you configure channel to support notices ("/mode #yourchannelname -n" or "/msg chanserv set mlock -n")23:39
wikingcould u plz?23:39
@sonney2kwiking, long done23:40
wikingConfigure your IRC channel to allow external messages (/mode #yourchannelname -n).23:40
@sonney2kwiking, that is the same thing23:40
@sonney2konly chanserv is doing this kind of stuff on freenode23:41
-GitHub153:#shogun- [shogun] none pushed 3 new commits to master:
-GitHub153:#shogun- shogun/master 046a89c Sergey Lisitsyn: Fixed ifdefs in SGBase to avoid failures with all but perl interfaces23:41
-GitHub153:#shogun- shogun/master 5f36e99 Sergey Lisitsyn: Removed wrong free of sgvector field of LinearKernel23:41
-GitHub153:#shogun- shogun/master 4e01cd0 Viktor Gal: Use SG_CALLOC instead of calloc Just to fix some of the static analyzer reports23:41
blackburnI told you sooooo23:41
wikingsonney2k: i guess it was your other account or?23:41
blackburnwiking: and then table is flying?23:41
@sonney2kwiking, show me your config23:42
@sonney2kwiking, no23:42
wikingsonney2k: just followed the23:42
wikingFor freenode, try the following:23:42
wiking# server: irc.freenode.net23:42
wiking# port: 666723:42
wiking# room: #yourroom23:42
wiking# message_without_join: checked23:42
wiking# long_url: checked23:42
wiking# notice: checked23:42
wiking# active: checked23:42
wiking# NOTE: Ensure you enable notice support (see above)23:42
-shoGitter:#shogun- [shogun] none pushed 3 new commits to master:
-shoGitter:#shogun- shogun/master 046a89c Sergey Lisitsyn: Fixed ifdefs in SGBase to avoid failures with all but perl interfaces23:43
-shoGitter:#shogun- shogun/master 5f36e99 Sergey Lisitsyn: Removed wrong free of sgvector field of LinearKernel23:43
-shoGitter:#shogun- shogun/master 4e01cd0 Viktor Gal: Use SG_CALLOC instead of calloc Just to fix some of the static analyzer reports23:43
wikingheheh even the nick part works23:43
@sonney2kwiking, no it is not my account because email notifications, cia, travis etc stuff is in there23:43
wikingah ok23:43
wikingtry it then with that23:43
wikingas u see it works23:43
wikinglet me turn it off23:43
blackburnwiking: where to set it up?23:43
wikingblackburn: admin -> service hooks -> IRC23:44
blackburnfound thanks23:44
wikingsonney2k: and then just push the test hook23:44
wikingit takes a about 10-20 seconds23:44
blackburnlets try23:44
wikingto get on irc23:44
blackburnwiking: how do you call it to announce again?23:45
wikingblackburn: test hook23:45
wikingafter you've updated the settings23:45
wikingyou'll have a test hook button23:45
-!- GitHub96 [] has joined #shogun23:45
-GitHub96:#shogun- [shogun] none pushed 3 new commits to master:
-GitHub96:#shogun- shogun/master 0d5f851 ptizoom: forced local shogun library and module path in tests/integration/perl_modular.23:45
-GitHub96:#shogun- shogun/master 6b6d83b Sergey Lisitsyn: Marked methods in director structured model as virtual23:45
-GitHub96:#shogun- shogun/master bbc4a3b Soeren Sonnenburg: Merge pull request #805 from ptizoom/perl_121026...23:45
-!- GitHub96 [] has left #shogun []23:45
-EDRTbot:#shogun- [edrt] none pushed 3 new commits to master:
-EDRTbot:#shogun- edrt/master d7797e7 Sergey Lisitsyn: Big bang23:45
-EDRTbot:#shogun- edrt/master ca8ca4a Sergey Lisitsyn: Towards proof of concept23:45
-EDRTbot:#shogun- edrt/master f3daae1 Sergey Lisitsyn: One more step towards prototype of KLLE23:45
blackburnwiking: haha23:45
wikingblackburn: :DDD23:45
heikosonney2k, amazing stuff :), but lots of programming work :D23:46
wikingsonney2k: works? :)23:46
heikosonney2k how to the $ -equivalent in mathjax?23:46
wikingsonney2k: let's call the bot Bartender :)23:46
-shogungit:#shogun- [shogun] none pushed 3 new commits to master:
-shogungit:#shogun- shogun/master 0d5f851 ptizoom: forced local shogun library and module path in tests/integration/perl_modular.23:47
-shogungit:#shogun- shogun/master 6b6d83b Sergey Lisitsyn: Marked methods in director structured model as virtual23:47
-shogungit:#shogun- shogun/master bbc4a3b Soeren Sonnenburg: Merge pull request #805 from ptizoom/perl_121026...23:47
wikingtravis already knows the drill :P23:48
wikinghe always serves another round on blackburn's tap :D23:48
@sonney2kwiking, alright it works now. the reason why it did not work was that the length of the nick has strict limitations23:49
wikingsonney2k: hehehe u gave like a 1024 chars long nick?:)23:49
heikoah got it23:49
@sonney2kwiking, nope around 1023:49
wikingsonney2k: :>23:49
wikingnone the less23:49
wikingthanx a lot23:49
blackburnsonney2k: was it DieShogunAlarmgeber?23:51
wikingblackburn: :D23:51
@sonney2kblackburn, of course sth like that - you know german have these long words like eierlegendewollmilchsau23:51
blackburnsonney2k: the password you gave me was like that as well23:52
wikingsonney2k: this d3 is pretty crazy23:53
@sonney2kblackburn, like HosenRunterUnterHosen :D23:54
@sonney2kwiking, yeah I love it. We need to craft all demos utilizing d3js - so we can have them live on the website23:54
wikingsonney2k: i suppose u can feed it with json data or?23:55
@sonney2kwiking, csv or cmdline or table or ...23:55
@sonney2kcmdline == as arguments23:55
@sonney2kseems like one can do anything with it23:55
wikingwe can generate json right?23:55
@sonney2kwe are python23:56
wikingnever tried it but i know that shogun depends on libjson :)23:56
wikingsonney2k: no i'm not  :(23:56
wikingsonney2k: fuckshit class inheritance doesn't let me python23:56
@sonney2kwiking, what are you trying to do?23:56
wikingwell as said earlier23:56
wikingif i want to be able to do latent svm fully from python23:57
wikingi would need to be able to define my custom classes (derived from CData)23:57
-EDRTbot:#shogun- [edrt] lisitsyn pushed 1 new commit to master:
-EDRTbot:#shogun- edrt/master 92c4ea0 Sergey Lisitsyn: Added REDSVD-like eigensolver routine (not tested yet)23:57
wikingand then i set these in my custom LatentModel (defined in python as well)23:57
wikingthis works fine23:57
wikingbut then23:57
@sonney2kblackburn, could you please stop the edrtbot logging?23:57
blackburnI will kill myself if it works faster than arpack23:58
wikingblackburn: :DDD23:58
blackburnsonney2k: why? I can23:58
wikingsonney2k: so but then when i try to get back the CData in LatentModel23:58
wikingi only get CData*23:58
wikingi cannot get back the class that i've defined in python23:58
wikingand there's no way i can 'cast' it to the right format23:58
wikingor at least i couldn't find a way23:59
@sonney2kblackburn, because of too much noise on the channel23:59
wiking(latentmodel stores CData* in a DynamicArray)23:59
blackburnsonney2k: I hardly commit once per day :D23:59
@sonney2kblackburn, this IRC thing is even more verbose than cia - not too nice :/23:59
--- Log closed Sun Oct 28 00:00:04 2012