Open in new window / Try shogun cloud
--- Log opened Wed Oct 24 00:00:00 2012
--- Day changed Wed Oct 24 2012
n4nd0atm I am checking the reference Cheng sent to the mailing list00:00
n4nd0wiking: do you know something about image segmentation using something similar to our so-svm?00:00
wikingn4nd0: heheheh00:00
wikingn4nd0: yeah i've checked that paper as well00:00
wikingi've seen some paper afaik00:01
wikingbut i have no fucking clue about img segmentation on the other hand00:01
wikingi know graph cuts :)00:01
wikingand that's it :>00:01
n4nd0hehe ok :)00:01
wikingfing help00:16
wikingi mean hell00:16
wikingwho the fuck creates big-endian formated images00:16
shogun-buildbotbuild #125 of nightly_none is complete: Failure [failed compile]  Build details are at
shogun-buildbotbuild #142 of nightly_default is complete: Failure [failed test]  Build details are at
-!- n4nd0 [] has quit [Ping timeout: 244 seconds]05:21
-!- n4nd0 [] has joined #shogun05:57
@sonney2kn4nd0, up early?06:20
n4nd0sonney2k: yeah! :)06:21
@sonney2kn4nd0, why that?06:21
n4nd0sonney2k: I am trying to find out what can I prepare using the SO framework for Christian's talk06:22
@sonney2kn4nd0, good excuse :D06:24
-!- n4nd0 [] has quit [Quit: leaving]07:03
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]08:22
-!- ptizoom [~christian@] has joined #shogun08:51
-!- blackburn1 [~blackburn@] has quit [Quit: Leaving.]09:57
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has joined #shogun10:28
wikingblackburn: any clearcases today? :D11:00
blackburnwiking: SURE11:00
blackburnmy life is empty without CC11:00
wikingblackburn: btw how do u like travis' message ?:) specially designed it for u11:00
blackburnyeah I like it11:01
wikingls -la11:06
wikingwrong window11:06
blackburngood you didn't wrongmessaged sth like 'cd porno/hard/german'11:09
blackburnwiking: what are you working on lately? a lot of sparse stuff, where is it used?12:17
-!- n4nd0 [] has joined #shogun13:10
wikingblackburn: i'd like to use it13:16
wikingblackburn: in my so-svm/latent sosvm implementation13:16
blackburnwiking: I just wojnder what is the problem you solve13:16
blackburnI mean sparse presentations are not so often used in case of local descriptors or anything like that13:16
wikingah malignant cell detection in xray chest images13:16
wikingbut it's not about the descriptors13:17
wikingit's about the representation of PSI features13:17
wikingin so-svm and latent sosvm13:17
wikinga lot of times you have like [0,...,0]_n [nnz]_m [0,...,0]_k like spi vectors13:18
wikingso you have zero padding before and after the nnz values13:18
wikingin the generated psi vector13:18
blackburnI see13:18
wikingso i want to be able to represent psi both as dense and sparse vectors13:19
wikingand for that i have this dotfeatures story13:19
wikingbut it's a pain in the ass13:19
blackburnsonney2k: it seems that SLEP problem is finally solved13:32
@sonney2kn4nd0: blackburn
n4nd0oh your widget application is ready14:45
@sonney2kthe calendar arrived14:45
n4nd0only 2011 visible so far?14:45
n4nd0sonney2k: any new about the doc sprint, are we applying finally?14:47
@sonney2kugh bug :-!14:47
@sonney2kn4nd0: sure ! we are14:48
n4nd0sonney2k: :)14:48
n4nd0so you take care of the application?14:48
@sonney2kn4nd0: btw 2012 is in chatlogs but at the very bottom.14:50
@sonney2kyeah I have to do it14:50
@sonney2kmaybe we should reverse the order14:51
n4nd0oh yes 2012 is there14:52
n4nd0I scrolled down too fast :S sorry14:52
n4nd0probably from newest date to oldest is more useful, I think14:53
@sonney2kptizoom: did you checkout git master?15:01
@sonney2kn4nd0, now it is reversely ordered
n4nd0wiking: ping15:18
n4nd0wiking: do you happen to know why I get this error ERROR MSK_RES_ERR_SPACE - 'Out of space.' in a problem that I could solve before?15:18
n4nd0wiking: I have just downloaded a new license in case it was that15:19
n4nd0but the error remains15:19
n4nd0sonney2k: nice!15:20
@sonney2kblackburn, the website is still confusing to me15:29
@sonney2kI think we should move
@sonney2kto the dev page15:30
@sonney2kblackburn, and look here
@sonney2knothing on it15:30
blackburnI'll put my drunk photo out there15:30
@sonney2kmaybe we better remove it15:30
@sonney2kthis info is under contacts anyways15:31
blackburnnot really I'd like it to stay15:31
@sonney2kblackburn, ok then write some longer content about it15:31
blackburnwith our photos and things like15:32
@sonney2klike list of all developers with coordinates etc15:32
@sonney2kplease do15:32
@sonney2khighly needed15:32
blackburn'ich bin sergej ich liebe das vodka'15:32
blackburnnot now I am at job :D15:32
@sonney2klater then15:32
* sonney2k is moving contribute over to dev15:32
@sonney2kblackburn, I think the references should be some subpage of about15:42
@sonney2kit is not really helpful for users...15:42
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]15:45
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has joined #shogun15:46
blackburnBLUE SCREEN OF DEATH15:46
blackburn<3 shindows15:47
@sonney2kn4nd0, blackburn
@sonney2kis this better?15:57
@sonney2k(as in more useful?)15:57
@sonney2kquestion is where should the 'we participate in gsoc2012' from go?15:58
blackburnsonney2k: I am ok with place it is now15:59
blackburnit is a good advertisement :D15:59
@sonney2kno it is basically hidden too deep16:00
@sonney2khard to find...16:00
n4nd0sonney2k: maybe we should include the HMSVM toolbox in acknowledgements?16:01
@sonney2kn4nd0, please do it16:02
@sonney2kactually also the GP toolbox16:02
@sonney2kstill a long way to go...16:02
@sonney2kthe dev page becomes useful16:02
@sonney2kbtw, I realize we have no explicit download page16:03
@sonney2kmaybe we should rename news to something else16:03
n4nd0sonney2k: to login and after edit?16:05
ptizoomsonney2k: I have not rebased for a while....16:06
@sonney2kptizoom, well don't rebase but just check it out16:07
ptizoom sonney2k: do I need to?16:07
blackburnich mochte schlafen16:10
ptizoomihren Sie erm?den16:16
@sonney2kn4nd0, admin/ sure16:34
n4nd0sonney2k: contributions updated16:34
@sonney2kptizoom, please test a clean git master and then try ./configure --interfaces=perl_modular16:34
n4nd0sonney2k: yeah ... I was missing the slash at the end, blackburn pointed it out :)16:34
ptizoomsonney2k: I see...16:35
n4nd0blackburn: what about the chef port?16:35
@sonney2kn4nd0, could you please add
@sonney2kare GP stuff is based on theirs16:35
n4nd0I just closed the window but yes I can do that16:36
@sonney2kplease do16:37
@sonney2kptizoom, we should get that to work in master16:37
@sonney2kthen we can add examples etc from your stuff16:37
n4nd0sonney2k: done16:38
n4nd0I am adding the matlab DR toolbox too16:40
ptizoomsonney2k: I have not much to test thy perl_modular, excepted my humoungus  batch  I put in tests/integration/pdl_modular16:40
@sonney2kptizoom, well test if it compiles!16:41
@sonney2kn4nd0, thx16:41
@sonney2kptizoom, then some example - if that works we can continue16:41
blackburnn4nd0: chef is a long run project sure16:43
blackburnwe are in design phase16:43
blackburnwe've got to estimate impact and regressions16:43
blackburni.e. how much money we can earn from that :D16:44
n4nd0need a break, see you later!16:46
-!- n4nd0 [] has quit [Quit: leaving]16:46
-!- heiko [] has joined #shogun16:59
-!- heiko [] has left #shogun []16:59
ptizoomsonney2k:...Checking for bzip2 ... yes17:19
ptizoomChecking for lzma ... no17:19
ptizoomChecking for Posix thread spinlock support ... yes17:19
ptizoomChecking for libshogun ... yes (internal in source tree)17:19
ptizoomChecking for Perl ... 5.14.217:19
ptizoomChecking for Perl ExtUtils::Embed ... yes17:19
ptizoomChecking for PDL ... yes17:19
ptizoomChecking for SWIG ... 2.0.717:19
ptizoomso far so good...17:19
@sonney2kptizoom, compile!17:21
ptizoomI have only a...celeron 2GHz (chipest...)17:24
ptizoomyou have warnings...17:25
ptizoomfor ShiftDataGenerator.cpp.o17:26
ptizoomi am disappointed17:28
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has quit [Quit: Page closed]17:30
ptizoomyet other warnings in sg_print_functions.cpp.o17:34
ptizoomaint working modshogun_wrap.cxx: In function 'pdl* if_piddle(SV*)':17:37
ptizoommodshogun_wrap.cxx:3568:22: error: 'PDL' was not declared in this scope17:37
ptizoomit shall had been defined like Core* PDL; in modshogun_wrap.cxx17:48
ptizoomlike in src/interfaces/modular/SGBase.i,17:49
ptizoompatched ...with #if  defined (SWIGPERL) //&& HAVE_PDL17:51
ptizoom+%header %{17:51
ptizoom+  SV* CoreSV;17:51
ptizoom+  Core* PDL;17:51
ptizoom+%init %{17:51
ptizoom+  //PTZ120930 boot PDL, load PDL stuff??? and define a PDL17:51
ptizoom+  //pl_require('PDL');17:51
ptizoom+  //check Core.xs //Core* PDL_p = pdl__Core_get_Core();17:51
ptizoom+  //PDL_COMMENT("Get pointer to structure of core shared C routines")17:51
ptizoom+  //PDL_COMMENT("make sure PDL::Core is loaded")17:51
ptizoom+  perl_require_pv("PDL::Core");17:51
ptizoom+  CoreSV = perl_get_sv("PDL::SHARE",FALSE);17:51
ptizoom+  //  PDL_COMMENT("SV* value")17:51
ptizoom+  if (CoreSV == NULL)17:51
ptizoom+    Perl_croak(aTHX_ "Can't load PDL::Core module");17:51
ptizoom+  PDL = INT2PTR(Core*, SvIV( CoreSV ));17:51
ptizoom+  //  PDL_COMMENT("Core* value")17:51
ptizoomsonney2k: have you got more patches in you bag or do you want me to carry on from there?17:53
ptizoom sonney2k:by the way all these //, #if 0 /**/ shall be zapped on the release branch!17:57
-!- Netsplit *.net <-> *.split quits: wiking18:08
-!- wiking [] has joined #shogun18:08
-!- wiking [] has quit [Changing host]18:09
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun18:09
wikingmmm splitting...?18:09
wikingsonney2k: ping18:28
-!- zxtx [] has quit [Ping timeout: 245 seconds]18:39
-!- blackburn [~blackburn@] has joined #shogun18:59
-!- n4nd0 [] has joined #shogun19:38
-!- zxtx [] has joined #shogun19:48
blackburnn4nd0: got any ideas for example?20:02
n4nd0blackburn: yep20:02
n4nd0I am working hard on it since this afternoon20:02
blackburncan I help you?20:03
n4nd0let's see if I get sth cool in time20:03
n4nd0mmm I don't know20:03
n4nd0brb in a minute20:03
n4nd0blackburn: I tell you the idea20:05
n4nd0so in this pattern recognition course I finished last week we developed an isolated word recognizer using HMMs20:06
n4nd0let's say you want to recognize numbers from 1-520:06
n4nd0then you train 5 HMMs, each HMM models a number (the word)20:06
n4nd0for new examples, you just run an algorithm in every HMM and recognize with the HMM that gives highest score20:07
n4nd0I aim at doing exactly the same thing, but training HM-SVMs instead of HMMs20:07
blackburncan you describe example of data?20:07
blackburnso it is a text?20:07
n4nd0it is speech20:07
n4nd0you record your voice saying 'one'20:08
n4nd0saying 'two'20:08
n4nd0do you see what I mean?20:08
blackburnyes probably20:08
n4nd0I just need to implement a new StateModel20:09
n4nd0so far we've got in shogun the twoState20:09
n4nd0for speech recognition you use other kind of state models20:09
blackburnI see20:09
n4nd0blackburn: do you think it sounds ok?20:11
n4nd0I think it is something I should be able to do for tomorrow and prepare a couple of slides with fancy pictures20:12
-!- n4nd0 is now known as n4nd0_away20:17
blackburnwould be nice20:37
-!- zxtx [] has quit [Ping timeout: 272 seconds]20:55
-!- n4nd0_away is now known as n4nd020:56
@sonney2kptizoom: is that stuff already in Sgbase?20:57
ptizoomsonney2k: in my "pull-request" branch...!21:34
@sonney2kptizoom, ahh in SGBase.i21:47
@sonney2khmmhh quite cluttered - lemme try21:47
n4nd0sonney2k: can I ask you a question regarding PLiFs?22:20
@sonney2kI have to enter the airplane now22:21
n4nd0ups ok22:21
n4nd0good flight then22:21
shogun-buildbotbuild #579 of deb3 - modular_interfaces is complete: Failure [failed compile python_modular]  Build details are at  blamelist: Soeren Sonnenburg <>22:28
shogun-buildbotbuild #470 of deb2 - static_interfaces is complete: Failure [failed test libshogun]  Build details are at  blamelist: Soeren Sonnenburg <>22:31
shogun-buildbotbuild #464 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>22:33
shogun-buildbotbuild #11 of ubu1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>22:43
shogun-buildbotbuild #61 of rpm1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>22:46
-!- zxtx [] has joined #shogun22:48
ptizoomsonney2k:the perl stuff is quite an Entwurf, and needs nice reduction... when it works23:04
-!- zxtx [] has quit [Ping timeout: 244 seconds]23:19
--- Log closed Thu Oct 25 00:00:17 2012