Open in new window / Try shogun cloud
--- Log opened Sun Oct 28 00:00:04 2012
--- Day changed Sun Oct 28 2012
@sonney2kwiking, I see00:00
wikingand in c++ i return CData* from the dynarray00:00
wikingbut if i cannot do this casting00:00
wikingthen there's no way u can use this from python00:00
blackburnunsolvable shit00:00
wiking:(00:00
@sonney2kwiking, so you need to implement multiple classes right?00:00
wikingyes00:00
@sonney2kand one depends on the extended other one00:00
@sonney2kyeah - no idea00:00
@sonney2kask on the swig ml00:01
wikingand since it's always application dependent00:01
wikingi cannot even like predefine a set of classes in c++00:01
@sonney2ksolution would be to have one class doing both :D00:01
wikinghaha00:01
wikingoverkill00:01
wiking:>00:01
@sonney2kwiking, well ask on the swig ML!00:02
blackburnsonney2k: I will patch PCA soon to use redsvd00:06
blackburnimpressive thing00:06
blackburndecomposition of 10k x 10k sparse matrix in 1.5 seconds00:06
@sonney2kblackburn, what is redsvd?00:08
blackburnhttp://code.google.com/p/redsvd/wiki/English sonney2k00:08
@sonney2kimpressive and yet another dependency :/00:10
blackburnsonney2k: all code in .hpp00:10
blackburn:D00:10
blackburn20 lines lol00:10
@sonney2kbut I guess we just stop counting :D00:10
@sonney2k20 lines only ???00:10
@sonney2khow come?00:10
blackburnsonney2k: http://code.google.com/p/redsvd/source/browse/trunk/src/redsvd.hpp00:11
heikosonney2k, I now added a feed for all entries on the site00:11
heikohttp://herrstrathmann.de/?format=feed&type=rss00:11
blackburnsonney2k: all the method are in lines 50-8600:11
blackburnsonney2k: I think all nice stuff is pretty compact00:13
blackburnliblinear is small yet efficient as well00:13
@sonney2knot that small00:16
@sonney2kheiko, ok I am adding this00:16
blackburnsonney2k: liblinear you mean? multiclass is very compact00:17
@sonney2kheiko, done00:18
-!- heiko1 [~heiko@host86-174-151-159.range86-174.btcentralplus.com] has joined #shogun01:25
-!- heiko [~heiko@host86-176-2-92.range86-176.btcentralplus.com] has quit [Ping timeout: 240 seconds]01:27
wiking2+1 = 2 tonight :D02:03
-!- heiko1 [~heiko@host86-174-151-159.range86-174.btcentralplus.com] has quit [Quit: Leaving.]02:25
-!- ptizoom [~christian@79-71-87-88.dynamic.dsl.as9105.com] has quit [Quit: Ex-Chat]03:18
-!- ptizoom_ [~christian@79-71-87-88.dynamic.dsl.as9105.com] has joined #shogun03:18
shogun-buildbotbuild #148 of nightly_default is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/14804:07
-!- ptizoom__ [~christian@79-71-87-88.dynamic.dsl.as9105.com] has joined #shogun07:08
-!- ptizoom_ [~christian@79-71-87-88.dynamic.dsl.as9105.com] has quit [Read error: Connection reset by peer]07:08
-!- Netsplit *.net <-> *.split quits: audy, naywhayare, shogun-buildbot07:43
-!- Netsplit over, joins: audy, naywhayare, shogun-buildbot07:48
-!- ptizoom__ is now known as ptizoom07:55
-!- Netsplit *.net <-> *.split quits: audy, naywhayare, shogun-buildbot08:43
-!- Netsplit over, joins: audy, naywhayare, shogun-buildbot09:30
-!- Netsplit *.net <-> *.split quits: audy, naywhayare, shogun-buildbot10:34
-!- Netsplit over, joins: audy, naywhayare, shogun-buildbot10:38
-!- wiking [~wiking@info2k1.hu] has quit [Changing host]11:38
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun11:38
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 252 seconds]12:27
-!- blackburn1 [~blackburn@188.168.3.27] has joined #shogun12:35
-!- blackburn [~blackburn@188.168.4.221] has quit [Ping timeout: 245 seconds]12:38
-!- gsomix [~gsomix@188.168.3.27] has joined #shogun14:25
-!- heiko [~heiko@host86-177-179-234.range86-177.btcentralplus.com] has joined #shogun14:39
-!- heiko [~heiko@host86-177-179-234.range86-177.btcentralplus.com] has quit [Quit: Leaving.]14:44
-!- gsomix [~gsomix@188.168.3.27] has quit [Ping timeout: 245 seconds]15:13
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun15:30
-!- gsomix [~gsomix@188.168.3.27] has joined #shogun15:30
wikingping16:55
wikingblackburn1: yo here?16:55
blackburn1wiking: yeah17:01
-!- blackburn1 is now known as blackburn17:01
wikingso what's with that new lib? :)17:08
blackburnwiking: which one?17:11
wikingthat u were testing yesterday17:12
blackburnhmm testing?17:13
wikingwell yeah17:20
wikingthat svd stuff17:20
blackburnahhhh17:21
blackburndidn't manage to test it yet17:21
blackburnwiking: works for gaussian so far, will test on swissroll just in a few minutes I hope17:26
blackburnwiking: no it doesn't work :D17:59
wiking:D18:01
wiking:(18:01
-shogungit:#shogun- [shogun] karlnapf opened pull request #806: added new perform_method, which (master...master) https://github.com/shogun-toolbox/shogun/pull/80618:02
-shogungit:#shogun- [shogun] karlnapf pushed 2 new commits to master: https://github.com/shogun-toolbox/shogun/compare/bbc4a3bbb433...1629d82d760618:02
-shogungit:#shogun- shogun/master 12265ad Heiko Strathmann: added new perform_test method which makes testing more efficient in certain cases and is a wrapper otherwise18:02
-shogungit:#shogun- shogun/master 1629d82 Heiko Strathmann: Merge pull request #806 from karlnapf/master...18:02
blackburnwiking: I managed to develop a gaussian generator so far :D18:04
blackburninstead of dimensionality reduction18:04
-!- gsomix [~gsomix@188.168.3.27] has quit [Ping timeout: 245 seconds]18:26
-!- gsomix [~gsomix@188.168.3.27] has joined #shogun18:31
-!- gsomix [~gsomix@188.168.3.27] has quit [Ping timeout: 245 seconds]18:41
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun19:09
n4nd0sonney2k: hi, I think it would be good if shogun-toolbox.org/irclogs redirects to this page/contact/irclogs (the calendar) instead of the current one19:10
n4nd0sonney2k: what do you think?19:10
blackburnn4nd0: just fix that ;)19:17
n4nd0blackburn: I will probably screw it up if I change something of the web :P19:18
blackburnI hope you won't :)19:18
blackburnn4nd0: my prototype works like gaussian generator19:33
blackburn:D19:33
n4nd0blackburn: what does that mean? hehe19:33
n4nd0oh19:33
n4nd0a bit random?19:33
n4nd0:O19:34
blackburnyeah it has an answer for the question of universe19:34
blackburngaussian blob that is19:34
blackburnn4nd0: I localized the problem though19:36
n4nd0good new19:36
blackburneigen brings some pain in ass as well19:48
blackburngaussian generator in da house19:52
n4nd0hehe19:53
blackburnn4nd0: oh! I've got a cross scatter plot19:54
blackburnmay be he wants me to become a christian19:54
n4nd0blackburn: may I see?19:55
n4nd0hehe19:55
blackburnn4nd0: https://dl.dropbox.com/u/10139213/shogun/fuckthisshit.png19:55
blackburnwhat can I say is that it is not a swissroll embedding yet19:56
blackburn:D19:56
n4nd0haha19:58
n4nd0definitely it is not19:58
blackburnI expect something is wrong in eigendecomposion as I fixed weights computation19:58
blackburnprobably it is a bad idea to start with experimental method19:58
blackburnhmmm I must have crazy lapack skills :D20:04
blackburnwrapper in 6 minutes20:04
blackburnshould write a book20:05
-!- heiko [~heiko@host86-177-179-234.range86-177.btcentralplus.com] has joined #shogun20:05
heikosonney2k, blackburn, around?20:05
blackburnich bin around20:05
heikohallo blackburn, wie gehts? :)20:06
blackburnI do know what is the question but no idea how to answer20:06
blackburn:D20:06
heikoblackburn, if I ask a kernel machine to give me its kernel via get_kernel(), and I do this in python, AND I know that this kernel is a Gaussian kernel, how can I call get_width() on it? since get_kernel returns type of CKernel, not CGaussianKernel20:07
blackburnfor some reason I know ich bin muede20:07
heikoblackburn, this was "How are you?" :)20:07
blackburnheiko: yes, I know20:07
blackburnbut no german analogue of 'I'm fine' in my mind20:07
blackburn:D20:07
heikoah , sorry, I see20:07
wikingheiko: is ther an obtain_from_generic() in CGaussianKernel20:07
heikoblackburn: "Gut, Danke!" "I am fine, thanks"20:07
wikingheiko: if not then define one20:07
wikingand call that from your python code20:07
blackburnheiko: you are in wiking 22 situation20:07
heikookay, cool this is exactly I wanted to know20:08
wikingblackburn: but he can fix it :)20:08
blackburnbut wiking 22 is even more tough20:08
wikingblackburn: as he has his class defined in c++ ;P20:08
heikoblackburn, just wasnt sure whether to use copy constructor or separate method20:08
blackburnit is just impossibru20:08
blackburnheiko: you have to unfortunately20:08
blackburnheiko: stupid stupid python20:08
heikoso copy constructor for CGaussianKernel AND obtain from generic, right?20:08
wikingyou don't need that20:09
heikoblackburn, thanks20:09
wikingheiko: just define a static function20:09
wikingobtain_from_generic20:09
heikowiking, static function?20:09
wikingand you're good to go20:09
wikingyeah check out other classes how they deal with it20:09
heikowiking, ok, thanks, what is an example?20:09
blackburnn4nd0: I've got another cross now with lapack20:10
heikowiking, I just saw dense features20:10
blackburnone more patch and I will see jesus20:10
heikobut there it is not static20:10
wikingheiko: src/shogun/structure/MulticlassSOLabels.h20:10
heikouh20:11
heikoannoying,20:11
wikingheiko: you can do this with EKernelType20:11
heikowiking, blackburn, we need this for every class, is there some way of not having to implement this all the time? :)20:11
wikingheiko: get_kernel_type and then do the typecasting20:11
blackburnheiko: I'd like to have one20:11
blackburn:D20:11
heikoI am lazy20:12
heikothis is needed for all models selection classes20:12
heikowhat about adding this to CKernel?20:12
blackburnit is out of OOP to be honest20:12
wikingheiko: well that won't help u20:12
heikowiking, and then using copy constructors20:13
wikingas u always have to return the right class type20:13
heikoyeh20:13
wikingheiko: and what's going to be the definition of that function20:13
wiking? :)20:13
wikingwhat will it return, CKernel* ?:)20:13
wikingbecause then you are at square one20:13
heikowiking, yeah ........20:13
heikodamn :)20:13
heikook, then, people will have to ask for that everytime they need it20:14
wikingheiko: you'll most probably have to define this everywhere20:14
wikingbut of course20:14
wikingyou can do a macro hack i think20:14
wikingso do this magic with macros20:14
wikingand then the macro will generate you the code for compilation20:14
wikingor templating ;)20:14
wikingyeah templating could work somehow :)20:15
heikowiking, yes, that might be possible, but is even more ugly :D20:20
heikowiking, blackburn, what about reference counting?20:24
blackburnheiko: what about20:24
blackburnref counting?20:24
blackburn:)20:24
heikothis is not done multiclasssolabels20:24
heikono sgref20:24
heikodont we need that20:24
blackburnhmm20:24
heikolike in get_kernel of kernel machine20:24
heikosince it produces a new reference to the object?20:25
wikingheiko: no it's just typecasting20:25
blackburnheiko: no just typecast for me20:25
wikingit doesn't do anything else20:25
heikowiking, but when I do it in python I have two references to the same object20:25
wikingbut since there's no way you can do typecasting in python20:25
wikingyou have to do this ugly hack20:25
heikook then, thanks guys20:26
wikingwhen u do get_kernel() or whatever that'll ++ the refcounter20:26
blackburnheiko: interesting point - can you check that?20:26
heikoblackburn, how?20:26
wikingafaik for a typecast u don't need an extra refcounter on it20:26
wikingi certainly don't see a point for it :P20:26
blackburnwiking: he means following:20:26
heikoI mean actually, an SG_REF would be needed, its needed whenever we produce a new reference to an object20:26
blackburnx = get_kernel() <- first appeareance of object20:27
blackburnblablakernel.obtain_from_shit(x) <- second one20:27
blackburnof the same object20:27
wikingblackburn: but that second one should not ++ the refcounter20:27
wikingas u essentially just casting there20:27
blackburnwiking: it doesn't but after these two lines we have two references20:27
blackburnto the same object20:27
blackburnbut +1 of ref counter only20:27
wikingand? :)20:28
wikingpersonally i wouldn't do it20:28
blackburnbut when they are going to be destructed20:28
heikoIf I now pass the first object to another place in python20:28
blackburnit is -2 for ref counter right?20:28
heikoand then the method is exited, then this object I passed on will be destroiyed20:28
heikoI mean nobody will do this anyway, but technically, we need another +1 whenever we produce a reference,20:29
heikoDoesnt hurt adding it, python will decrease it automatically20:29
wikingheiko: yes python20:30
wikingnot c++20:30
wikinghence it'll hurt some other code20:30
wikingbut if u fix it20:30
wikingthen go ahead20:30
wiking:)20:30
wikingi certainly valgrind it20:30
wikingas there's already there's some shit already with streamingfeatures20:30
heikobut in c++ the agreement is that whenever you get an object, the ref-count is increased20:30
heikosee get_kernel() etc20:30
heikoalso why should one use the method from c++ ?20:30
wikingheiko: it's cleaner :))))20:31
wikingheiko: and it does a typechecking20:31
wikingalthough of course20:31
heikoI would rather do a dynamic_cast20:31
wikingin c++ you could try dynamic_cast20:31
wikingyes indeed20:31
wikingbut sonney2k killed me for that one20:31
heikofor dynamic casts?20:31
heikowhy that?20:31
wikingwhen i did it in one of my code20:31
wikingyeah he made me remove it :>20:31
heikofor what reason?20:31
heikoActually, one doesnt even need a dynamic cast20:32
heikoa simple cast is ok20:32
wikingshame one cannot search in PRs20:32
wikingheiko: well dyncast at least does some checking20:32
heikosince the method will throw an error if the type is wrong anyway20:32
wikingand it'll throw an exception if u wanna do some nasty shit20:32
heikoone should never cast if one isnt sure about the type anyway20:32
heikoanyway, different topic :)20:32
-shogungit:#shogun- [shogun] karlnapf opened pull request #807: added obtain_from_generic to GaussianKernel (master...master) https://github.com/shogun-toolbox/shogun/pull/80720:37
blackburnn4nd0: fuck this shit, it works!20:53
wiking\o/20:55
wikingnotmine20:55
wiking:(20:55
blackburnwiking: heh20:55
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 240 seconds]20:56
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun21:11
heikoblackburn, still there?21:21
blackburnheiko: yes21:21
heikoblackburn, there is a memory bug in CCrossValidationMulticlassStorage21:21
heikoblackburn, related to pointers to SgVectors21:21
heikoI dont know about this, I think we shold totally avoid these21:21
heikothey cause massive trouble on the long run21:21
blackburnheiko: like arrays of sgvectors?21:22
heikoyes21:22
heikothese should not exist in my eyes21:22
blackburnhah what to use instead?21:22
heikowe have to think of something21:22
wikingstd::vector<SGVector<> >21:23
heikoif we continue like this, we will totally destroy any serialization21:23
wikingFTW21:23
heikoalso its so hard to debug when there is a mixture of auto-ref-counting and these arrays of SGVectors21:23
heikoblackburn, I mean usually you know how many vectors you want to have, so just ust a matrix21:23
blackburnheiko: SGVector<>[] would work too21:23
heikowhat about a matrix instead?21:23
blackburnit is not always possible21:24
heikoAnd we *definitely* need to think about a data structure that can dynamically handle SGVectors21:24
heikoblackburn, but its possible most of the time21:24
heikoe.g. in CCrossValidationMulticlassStorage21:24
blackburnheiko: std:: vector :D21:24
heikoyou know how many you store before21:24
heikoblackburn, no, so std:: things please21:24
heikothink of serialization21:24
heikowe *need* a new structure for this21:24
blackburnheiko: just copy a std::vector then21:25
heikoblackburn, what do you have against a matrix?21:25
heikowhats the difference to SGVector<>*21:25
heiko?21:25
blackburnheiko: nothing I am just unsure it is always possible21:25
heikoblackburn, its not - but only if we need dynamic structures21:26
heikoin this case SGVector<>* wont help anyway,21:26
heikoso we can always avoid it21:26
heikoand for the dynamic case, we need something new21:26
heikothat can handle SGReferenced data21:26
heikoalso in the multiclass labels, there we can replace everything by SGMatrices21:27
heikosince its not dynamic21:27
wikingi need a dynamically resizeable Matrix21:27
heikoThis way, we ensure that shoguns parameter framework still can handle everything21:27
wikingso if u come up with something then plz let me now21:27
wikingcurrently i do21:27
heikowiking, yes I need one too21:27
wikingstd::vector<SGVector<> >21:27
blackburnwiking: EIgen::MatrixXd21:28
blackburn:D21:28
heikowiking, please dont use this construction21:28
wikingheiko: hehehe yeah that's my solution :)21:28
wikingheiko: why not? :)21:28
wikingworksforme21:28
wiking:D21:28
wikingbut i know that it's not allowed in shogun21:28
wikinghence i haven't commited the code21:28
heikowiking, you will kill shogun :(21:28
heiko*poor shogun*21:28
wikingblackburn: ok so we are going with hard-linking to Eigen3? :)21:29
wikingi have nothing against it  :D21:29
blackburnwiking: you know I have nothing against anything ;D21:29
heikoblackburn, could you have a look in http://pastebin.com/fyCnxHxc ?21:30
heikothis reproduces the bug I mentioned21:30
blackburnheiko: man it is big, what is the bug?21:30
blackburn:D21:30
heikojust run with valgrind21:31
heikomemory bug in the CrossValidationMulticlassStorage21:31
blackburnheiko: can you line me up?21:31
heikoblackburn, private msgs got throttled :)21:32
heikothe bug happens in destructor of the class21:32
blackburnheiko: yeah I've got a message21:33
heikojust run valgrind and you will see21:33
blackburnheiko: I just talked to gsomix21:35
blackburnhe is 2 meters far away21:35
blackburnyou thing can be solved other way21:35
heikoblackburn, send him my greetings21:35
heikoblackburn, im curious :)21:35
blackburnpython_modular/swig_typemaps.i:82521:36
blackburngreetings in return heh21:36
blackburnheiko: that maps features to its actual type21:36
heikoah that is related to the obtain_from_generic thing right?21:37
blackburnyes21:37
blackburnnot that storage shit :D21:37
heikook, so you think the other problem can be solved using a similar concept?21:37
blackburnkernel problem should be solved21:38
heikoblackburn, yes, nice idea21:39
heikoblackburn, is gsomix interested in implementing that? :D21:40
blackburnheiko: not now but he could21:40
heikoblackburn, great then! :)21:40
heikoblackburn, about that storage shit21:41
blackburnyeah21:41
heikoI think we should change things asap21:41
heikootherwise, we will collect more and more21:41
heikoalso get rid of using std::vector whereever its possible21:41
heikoThe whole parameter framework becomes obsolete if we dont do this21:42
blackburnheiko: one example21:42
blackburnconfusion matrices21:42
blackburnhow to store array of confusion matrices?21:42
heikoblackburn, yes good point21:42
heikoWe need a structure for these SGReferenced objects21:42
blackburndynamic?21:42
heikowhere we pass by reference21:43
heikoyes21:43
blackburnwhy by reference?21:43
heikobecause that is the way the SG* are handled21:43
heikootherwise we could just derive from SGObject21:43
heikoand handle it like other shogun classes21:43
heikobut the whole point was not doing that21:43
blackburnehmm we can pass it by value with no problem21:43
heikoah21:43
heikosorry21:43
heikoI meant that21:43
heikobut not by pointer21:44
heikoand then we can make this structure serializable and we are done21:44
wikinghttp://www.welovetheherb.com/wp-content/uploads/2012/05/Fuck-This-Shit-Get-The-Bong.jpg21:45
wikingindee21:45
wikingd21:45
blackburnwiking: yes21:45
heikowiking :D21:46
heikobut seriously - this has to be changed, we will also get rid of debugging errors like the one in CCrossValidationMulticlassStorage21:46
wikingheiko: no worries i was talking 'bout my work21:46
wiking;)21:47
blackburnheiko: can you see anything wrong in that storage?21:49
heikoblackburn, I will take a look21:53
blackburnI am in  eigenfighting now :D21:54
wikingYo trabajo sin reposo21:54
wiking!!!21:54
blackburnwiking: chego?21:55
wikingfuckthisshittime21:55
wikingwith buena vista social club ;)21:55
wikinghow the fuck is it possible21:56
wikingthat i get the same PO21:56
wikingwith C=1 and C=15021:56
wikingdoes not compile21:56
heikoblackburn, m_fold_ROC_graphs is not initialized21:58
heikosetting it to NULL in the constructor solves the problem here21:58
blackburnheiko: i see22:00
heikoI will correct and push22:00
blackburnthanks22:00
heikoblackburn, any ideas for a SGReferenced structure?22:01
blackburnheiko: no not really we will imitate std::vector in the end :D22:01
heikoblackburn, what about somehow being able to use the existing data structures for that -- since they already work fine22:02
blackburnheiko: which?22:03
heikoCDynamicObject, CList, etc22:03
blackburndynamicobjectshit should work22:03
heikohuh? how ?22:04
heikoIt accepts CSGObject* only22:04
blackburnahh22:04
-shogungit:#shogun- [shogun] karlnapf opened pull request #808: fix for a memory error (master...master) https://github.com/shogun-toolbox/shogun/pull/80822:05
-shogungit:#shogun- [shogun] karlnapf pushed 2 new commits to master: https://github.com/shogun-toolbox/shogun/compare/b936a24f6bd0...3c4a075bdb8722:05
-shogungit:#shogun- shogun/master cd3115e Heiko Strathmann: fixed a memory error due to uninitialized array pointers22:05
-shogungit:#shogun- shogun/master 3c4a075 Heiko Strathmann: Merge pull request #808 from karlnapf/master...22:05
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]22:06
@sonney2kwhats up here?22:15
heikosonney2k, hi there :)22:15
@sonney2kwiking, I asked you to use obtain_from_generic instead of dynamic cast (it will catch error situations)...22:16
@sonney2kIIRC22:16
@sonney2kheiko, what are you trying to emulate?22:16
heikosonney2k, what do you mean?22:17
@sonney2kI think we are just fine with any SGVector[] or whatever as long as we don't need them as parameters / ms-parameters22:17
@sonney2kactually any data structure is fine as long as we don't need to serialize it / use it in MS22:17
heikosonney2k, indeed22:17
heikosonney2k, but actually many places where these are used are cases where they are parameters22:18
heikofor example multiclass labels22:18
@sonney2kI am wondering a bit whether gsomix managed to do some hackery to always return the correct type from python...22:18
heikocurrently completely non-serializable22:18
heikosonney2k, another downside in my eyes is that using SGVector[] is a step back to where we came from before SGVectors were around22:19
@sonney2kheiko, yeah but this case would have worked with SGMatrix w/o problems22:19
heikosonney2k, yes true, actually most places will work with matrix22:20
@sonney2kheiko, true22:20
@sonney2kwe of course have SGStringList22:20
heikosonney2k, what about if one needs multiple matrices22:20
@sonney2kthat is basically doing that22:20
@sonney2kyou can have vectors of arbitrary mixed lengths22:21
@sonney2kmultiple matrices - SGNDArray22:21
@sonney2kbut fixed length again22:21
heikowhich is not implemented anywhere22:21
@sonney2kheiko, you mean serialization is not but NDArray is22:21
@sonney2kheiko, serialization for SGNDArray is easy though (if you have it working for SGVector)22:22
heikosonney2k, true22:22
@sonney2kit is just a flat data structure like SGVector/Matrix22:22
@sonney2kheiko, we don't have stringlist serialization do we?22:23
heikosonney2, actually dynamic structures are not that much needed since one could just use a std:: structure and then write back to a fixed SG structure afterwards22:24
@sonney2kheiko, and SGSparseMatrix serialization?22:24
heikobut I am not aware of any case where a parameter is a dynamic thing22:24
@sonney2kwell a list of strings - but that currently is only used in features22:24
@sonney2kactually we might need to rename SGString to sth SGVariableLengthArrays or so22:25
heikoyes maybe22:25
@sonney2kSGStringList I mean22:25
heikoyes22:25
heikowould be nice, to make it more "visible"22:25
heikosonney2k, I am curious about this SGNDArray22:26
heikois this actually working?22:26
heikobecause then, transition from SGMatrix[] to a SG-parameter would be quite easy22:28
@sonney2kheiko, it is used in structured output stuff22:28
heikowow didnt know22:28
wikingsonney2k: as dyncast catch as well errors... or?22:28
@sonney2kheiko, well not the new one some old one22:29
@sonney2kwiking, no dyncast will not throw a shogun exception but just return NULL22:29
heikosonney2k, do you got a class name for that?22:29
wikingsonney2k: but nevermind i was just saying to heiko that dyncast is out of scope:>22:29
@sonney2kwiking, so you might get crashes / have to check manually22:29
@sonney2kheiko, for what?22:29
heikosonney2k, for the SO NGNDArray22:29
heikosonney2k, I use dynamic_cast sometimes exactly because of this nice thing that you can check for NULL22:30
wikinghahahah my machine give always +1 as prediction label :)22:31
blackburnwiking: positive guy22:31
wikingyeah amazing shit22:32
@sonney2kheiko, yeah but if you expect a certain type at a particular location in the code than there is no use - you will have to throw an exception anyway. that is why we wrote the obtain_from_generic * stuff22:33
@sonney2kwiking, congratulations - I bet I could have learned sth like that *even* faster22:34
wikingsonney2k: :>>>22:34
@sonney2kheiko,22:34
heikosonney2k, ok, well my need of this had a different motivation22:34
@sonney2kgrep SGNDArray shogun/ -rl --include='*.h'22:34
@sonney2kshogun/multiclass/QDA.h22:34
@sonney2kshogun/lib/SGNDArray.h22:34
@sonney2kshogun/structure/PlifMatrix.h22:34
@sonney2kshogun/structure/DynProg.h22:34
wikingsonney2k: yeah it took me 3000secs to learn this w :D22:35
@sonney2kheiko, so n4nd0 is using it too22:35
@sonney2kwiking, w might be ok just b not22:35
wikingsonney2k: nah i'm reading now the w from a file.... and the other code does better, so i think my port of the problem is buggy22:35
heikosonney2k, cool thanks22:35
blackburnLOL22:36
blackburnG = [ 44.7894 -3.80941  30.971922:36
blackburn-3.80941  24.3547 -24.103922:36
blackburn 30.9719 -24.1039  48.6117]22:36
blackburnbest gram matrix ever22:36
heikoblackburn, so all the SGVector[] can be replaced by SGMatrix and SGMatrix[] can be replaced by SGNArray, and then we can register the things as parameters again22:36
blackburnheiko: go for it :D22:37
heikoblackburn, this was more to keep in mind for new code :D22:37
heikolike new model selection storage classes22:37
blackburnI wish I didn't create that class22:38
blackburn:D22:38
heikoblackburn, lazy guy :D22:38
heikosonney2k, have I told you about my new job already?22:38
heikocant remember if you were there last time ... ?22:38
@sonney2kheiko, no what is it?22:40
@sonney2kheiko, btw SGSparseMatrix and SGStringList also need to be supported by serialization22:40
@sonney2kI think they are not yet22:40
heikosonney2k, I will work on GP classification at uni, and since the guys are keen on an implementation, I convinced them that I do that for shogun22:41
heikoits gonna be some fancy stuff22:41
heikowith advanced MCMC for redictive distribution integrals and hyperparameter learning22:41
heikoso I will extend the GP framework with all that stuff22:41
heikoand get paid for it :)22:41
heikosonney2k, oh and since its a collaboration with a company who needs that for some quality testing on welds, they will use shogun for it22:42
blackburnjust to check my mind22:42
blackburnlinear kernel can't be negative right?22:42
blackburn:D22:42
heikoblackburn, [-1, -1] * [1,1]22:43
blackburnehmm22:43
heikoor what do you mean?22:44
@sonney2kblackburn, no mercer kernel should be - though there are numerical issues sometimes22:44
blackburnso scalar product can be negative :D22:44
blackburnI was confused but it is ok right?22:44
@sonney2kheiko, excellent?22:44
blackburnsonney2k: yes should be posdef but can contain negative elements?22:44
@sonney2kblackburn, sure22:45
@sonney2kblackburn, not on diagonal though :D22:45
blackburnsonney2k: yeah they are norms22:45
@sonney2kexcept for floatingpoint issues22:45
@sonney2knever underestimate the power of floats22:46
* sonney2k argh I couldn't resist - another 4 discus fish are now in the tank22:54
blackburnsonney2k: haha22:55
@sonney2kMadeira-Merah Discus - http://easydiskus.de/madeira-merah/22:55
heikosonney2k, these cool pretty cool :)22:55
heikodidnt know you were into that22:55
@sonney2kI hope they survive22:55
@sonney2kif they all do this night - then good22:56
heikohehe wont they got eaten by their small fellows?22:56
@sonney2kheiko, what?22:56
heikothese small ones on the pictures22:56
@sonney2kheiko, ahh no babies rarely eat their parents :P22:57
@sonney2kbut mine are not yet that old22:57
heikoyour babies are not that old to eat you? :D22:57
@sonney2kdiscus fish can live pretty long - up to age 16 years(!)22:57
@sonney2kheiko, I hope so22:57
@sonney2kand actually that they never attempt that!22:57
@sonney2kthe ones here are < 1 year22:58
@sonney2kbut >12 cm already22:58
@sonney2knot yet pizza size but close to22:58
heikowow pretty big22:58
@sonney2kand strong ones too22:58
@sonney2konly issue is that each one carries different bacteria - so when one has bad luck everyone is dead in the end22:59
@sonney2k(by end of tomorrow)22:59
@sonney2kheiko, btw you are actually partially right - baby discus feed from the skin of their parents in the first days23:01
@sonney2kan parents change color to dark brown23:01
@sonney2kand babys go for everything dark23:01
heikosonney2k, hehe, cool guys you got there.23:09
@sonney2kheiko, how long will your new contract last?23:11
heikosonney2k, ~6 months23:11
@sonney2kcool23:12
heikoyeh, looking pretty forward to it23:12
@sonney2kheiko, wish we could generate such kind of jobs23:12
heikoIn general, to learn about GPs in detail23:12
heikosonney2k, the original job description was:23:12
@sonney2kheiko, ask oliver stegle about stuff - he really knows his way around (Jacobs mentor that is)23:13
heikocome up with GP for multiclass classification (they already had GP and MCMC in mind), implement sophisticated software (should be usable by company), and write paper23:13
heikosonney2k, cool thanks for the tip23:13
heikosonney2k, btw you forwarded my email to gunnar some time ago, remmeber?23:14
heikoPhilipp Drewe got back to me and we did something together23:14
@sonney2kI don't even remember23:15
heikosonney2k, well thanks anyway :)23:16
@sonney2kheiko, btw we need to start the billing stuff soon23:23
blackburnsonney2k: heiko: I am not going to make my docs in time :(23:24
@sonney2kblackburn, how do you know?23:24
blackburnsonney2k: will receive my foreign passport after 13th of november23:24
blackburnthey scheduled it already23:24
heikosonney2k, ok thats fine here23:25
blackburnsonney2k: -> less than 2 weeks to get american visa23:25
heikoblackburn 13 november? isnt that fine?23:25
heikoah, fuck you need a visa23:25
heikodamn :(23:25
blackburnyeah23:25
@sonney2kblackburn, well we can still try no?23:25
heikoI will fly to Germany next week to get an express passport (otherwise I also need visa)23:25
blackburnsonney2k: I don't mind I just let you know as I received a notification23:26
@sonney2kblackburn, erm the doc summit starts on dec 3 right?23:26
blackburnsonney2k: right23:26
@sonney2kso about 3 weeks23:27
@sonney2kfor visa23:27
blackburnnot clear, I will receive passport *after* 13th23:27
@sonney2kheiko, I guess I have to do the job anyways anyways23:27
@sonney2kblackburn, I see23:28
blackburnso 2 weeks for visa in best case23:28
heikosonney2k, just let me know, I am quite flexible with time currently since no deadlines for the first time in ages :)23:28
blackburnsonney2k: funny thing I'd have to go to moscow for an interview23:28
heikoblackburn, time you move to europa :)23:29
blackburnso would require 2 days out23:29
@sonney2kheiko, I forgot what I had to do last time but it could be that I do it and you will only notice once the money arrived :D23:30
heikosonney2k, ok then tell me before I accidentially spend it on stuff ;)23:31
@sonney2kheiko, I will make you swallow some gps to track you :D23:32
heikohehe :)23:32
heikono worries about that23:32
heikoI quit all my drug addictions :D23:33
wikingdocsdocsdocs23:33
wikingblackburn: let's just forge u a passport that doesn'tneed visa to the states :23:34
blackburnwiking: passport of united states you mean?23:34
blackburn:D23:34
blackburndo you know a guy to do that?23:34
wikingmmm guess there's always a guy :)))23:34
wikingu need like a low profile passport23:35
@sonney2kexcept that they will chop of your head if they recognize23:35
wikinglike a slovakian one23:35
wikingcouple of bottles of good vodka should fix u up with one :P23:36
heikosonney2k, btw what about this live pdf of the shogun-tutorial?23:36
wikingblackburn: once i've heard that a bunch of guys who were maintaining/upgrading the database of citizens in new zealand got into trouble: somebody from the authorities realised that there were more rows in the new table than in the old one :DD23:37
blackburnhahah23:37
wikingblackburn: they confiscated all team members' passport and nobody was allowed to leave the country until the investigation finished23:38
wikingthat was around xmas23:38
wikingso a collegue's brother couldn't come back for xmas holidays as he was part of the team :D23:38
@sonney2kwiking, nice23:38
wikingmust be funny to add yourself a row in that table :)))23:39
@sonney2kheiko, well we generate the doxygen doc only currently (but only if nightly_default works !)23:39
@sonney2kheiko, surely we could do the .pdf too23:39
heikosonney2k, is it hard to add the latex to the nightly build?23:39
wikingblackburn: u need to do something like this :)))23:39
wikingblackburn: insert into citizen :D23:40
blackburnwiking: I'm just searching a way to do an sql injection23:40
wikingblackburn: ahhahahaha23:40
wikingor like get maltese citizenship23:40
wikingi bet that they have low security going on :)))23:40
blackburnwiking: there was a case recently23:41
@sonney2kblackburn, if you are lucky the same people that administrate sony's servers are taking care :D23:41
wikingsonney2k: :D23:41
@sonney2kheiko, no - easy23:41
blackburnwiking: pope riding a bmw car with maltese plates bammed into a few cars23:41
heikosonney2k, would be awesome to have in my eyes - there could be a link on the shogun website23:41
@sonney2kheiko, btw did you twitter?23:41
blackburnhe was also drunk (nice addition23:41
heikosonney2k, yes23:41
wikingblackburn: :>>>23:41
@sonney2kheiko, cool :D23:42
wikingblackburn: which pope?23:42
blackburnwiking: not guilty in the end23:42
blackburnwiking: argh priest may be?23:42
wikingblackburn: ah ok :>23:42
blackburnwiking: you know I am stupid with owrds :D23:42
wikingpope is the head of the church :>23:42
blackburnwiking: so orthodox priest23:42
blackburnwiking: ??? in russian is a priest23:42
blackburnso got confused23:42
heikosonney2k, btw I am also not a twitter person ;) my girlfriend had to explain it to me23:42
@sonney2kheiko, I am totally ignorant of all blogging / twittering etc23:43
@sonney2kI hate doing publicity23:43
@sonney2kbut it is a necessary evil23:43
@sonney2ktakes me lot of time I cannot to reasonable things23:43
heikosonney2k, same here, but its quite handy for this scientific networking23:44
heikosonney2k, lol, I did not even do my webpage myself :D23:44
wikingsonney2k: can we add a RT rule to shogun's twitter acc?23:44
wikinglike if it sees #shogun then RT it? :)23:44
heikowiking, I just got an introduction into that, and these hashs #23:45
wikingok bad idea23:45
wikinghttps://twitter.com/search?q=%23shogun&src=typd23:45
wiking#shogun-toolbox maybe then :)23:45
heikowiking, phew, I forgot it this time :)23:45
heikoso  better not do that23:45
heikomayne shogun-toolbox23:46
@sonney2kwiking, luckily not :)23:46
wikingsonney2k: as i see that shoguntoolbox has some followers23:47
wikingand instead of trying to do some co-tweeting23:47
wikingjust retweet if some-of-us mentions #shogun-toolbox in their tweet or soemthing23:47
@sonney2kwiking, I haven't used twitter at all - so totally clueless. Just created that account and uploaded a picture. Now I am hoping that heiko, n4nd0, blackburn fill in some content23:48
-shogungit:#shogun- [shogun] sonney2k pushed 1 new commit to master: https://github.com/shogun-toolbox/shogun/commit/915af5cda5ab6c46a5cddb04fa993ff4b9651cc023:50
-shogungit:#shogun- shogun/master 915af5c Soeren Sonnenburg: add make tutorial target23:50
wikingsonney2k: ic23:54
wikingblackburn: twitbot? :)23:55
@sonney2kheiko, wiking btw if we manage to get in gsoc next year and you intend to mentor - I've met some projects who had ~4 people attending23:55
heikosonney2k, 4 people at once in one project you mean?23:55
wikingattending the summit?23:55
@sonney2kyes. only problem will be flight costs I guess23:55
heikoah the mentor summit?23:55
heikonice :)23:55
wikingbtw sonney2k dont u wanna get the travis fixed23:56
wikingso it tries all the interfaces?23:56
@sonney2kI am not sure how much google covers then... they did cover 2 hotel nights and IIRC up to 2k$ for flights23:56
@sonney2kwiking, you are the travis hero23:56
wikinghah23:56
wikingdonthavetime23:56
@sonney2kI have no clue about it (never even looked at it)23:56
@sonney2kwiking, and btw also the unit tests stuff23:57
wikingsomeobody should smartly define ENV variables23:57
blackburnsonney2k: how do they choose # of mentors to arrive?23:57
wikingsonney2k: yeah i've started it but i'm quite busy with all the latent coding/fixing23:57
wikingheiko: what's your twitter acc?23:57
@sonney2kwiking, if it helps blackburn creates a ticket and assigns it to you :D23:57
heikowiking, dont have23:57
wikingheiko: ic23:58
@sonney2kblackburn, they have the rule 2 persons per project23:58
wikingsonney2k: :D23:58
@sonney2kblackburn, and then there is a waiting list23:58
wikingsonney2k: have u closed the SA ticket? :)))))23:58
blackburnsonney2k: ahh23:58
@sonney2kso some projects might not want to send 223:58
wikingsonney2k: https://github.com/shogun-toolbox/shogun/issues/78923:59
wikingsonney2k: u can close it :)23:59
blackburnargh this week I am going to receive reject (or accept if I am lucky)23:59
wikingblackburn: newjob?23:59
blackburnyes23:59
wikingwhich co?23:59
blackburnwiking: http://corp.wesee.com/en/home/23:59
wikingthe one who made u to hack xerox? :D23:59
--- Log closed Mon Oct 29 00:00:00 2012