Open in new window / Try shogun cloud
--- Log opened Fri May 04 00:00:37 2012
-!- cronor [] has joined #shogun00:50
-!- gsomix [~gsomix@] has quit [Ping timeout: 265 seconds]01:00
-!- cronor [] has quit [Quit: cronor]01:48
-!- av3ngr [av3ngr@nat/redhat/x-arwstdbdsgyjdant] has joined #shogun03:19
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]03:22
-!- vikram360 [~vikram360@] has quit [Ping timeout: 245 seconds]03:33
-!- SeattleDad [322e753f@gateway/web/freenode/ip.] has joined #shogun05:28
SeattleDadI'm trying to install shogun for lua on Mac OSX05:29
SeattleDadI'm getting a problem when I do make install05:30
SeattleDadIt says:05:30
SeattleDadinstall -m755 modshogun.dylib "//usr/local/lib/lua/5.1"05:30
SeattleDadand then I get05:30
SeattleDadinstall: modshogun.dylib: No such file or directory05:30
SeattleDadI don't see this modshogun.dylib anywhere05:31
-!- vikram360 [~vikram360@] has joined #shogun05:58
-!- vikram360 [~vikram360@] has quit [Ping timeout: 260 seconds]06:12
-!- vikram360 [~vikram360@] has joined #shogun06:15
-!- vikram360 [~vikram360@] has quit [Ping timeout: 252 seconds]06:22
-!- vikram360 [~vikram360@] has joined #shogun06:23
-!- blackburn [~qdrgsm@] has quit [Ping timeout: 276 seconds]06:41
-!- cronor [] has joined #shogun07:58
-!- gsomix [~gsomix@] has joined #shogun08:18
gsomixhi all08:18
@sonney2kSeattleDad, which version of shogun?08:22
CIA-113shogun: Soeren Sonnenburg master * rf46645d / (3 files in 2 dirs): fix a few crashers related to sgvector transition -
-!- vikram360 [~vikram360@] has quit [Ping timeout: 252 seconds]08:48
-!- wiking [] has joined #shogun08:49
-!- wiking [] has quit [Changing host]08:49
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun08:49
sonne|work~~~~...has broken~~~like~~~~ooo09:02
-!- uricamic [~uricamic@2001:718:2:1634:f2de:f1ff:fecf:a6a5] has joined #shogun09:04
gsomixmorning :)09:05
wikingsonne|work: what happened?09:08
sonne|workwiking: I am singing09:09
sonne|workguess that must be some lack of sleep thing09:09
wikingahahahah :D09:09
wikingyeah that usually is a good indicator09:09
gsomixit's time to compile deadbeef player. I need music.09:10
gsomixsonne|work, how are you?09:10
sonne|workbut sgvector is making good progress09:13
sonne|workI am very happy with the result09:13
sonne|workit is soooooo much easier to use09:13
sonne|workwish we had that w/ other SGObjects09:15
-!- cronor [] has quit [Quit: cronor]09:16
-!- karlnapf [] has joined #shogun09:16
sonne|workkarlnapf: hey09:17
sonne|workkarlnapf: I was fixing a few sgvector related things in modelselection/xval09:17
sonne|workand I stumbled across your hack09:17
karlnapfsonney2k, which one? :)09:17
sonne|workSGVector<char> m_dontremember;09:17
karlnapfoh year, that wasnt actually mine09:18
sonne|workwhere you then later assign m_donot.vector / vlen09:18
karlnapfsomebody inserted that09:18
sonne|workhaha me then :D09:18
karlnapflol :)09:18
sonne|workproblem is that it really needs careful workarounds to work well with the new stuff09:18
sonne|work-> not nice09:18
karlnapfok, but this is avoidable isnt it?09:19
sonne|workkarlnapf: actually I remember you used SGVector<void> !09:19
karlnapflol, yes even worse ;)09:19
sonne|workyeah it should be09:19
sonne|workmaybe you have an idea if you look at it09:19
karlnapfok Ill check09:19
sonne|worksgvector fixes are relatively simple09:19
karlnapfhave you seen the email about appended features for group lasso?09:20
sonne|workusually related to some additional SG_FREE(v.vector)09:20
sonne|workdidn't have time for that yet09:20
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has joined #shogun09:21
karlnapfThe term "features" might be a fit fuzzy there, but maybe its worth investing some more time into the appended features09:21
blackburnhey there09:21
karlnapfgood morning honey :)09:21
blackburnkarlnapf: oh our relationship have got some progress this night I guess :D09:22
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]09:23
sonne|workkarlnapf: dont' get in my way - I am already engaged with blackburn!09:23
karlnapf:D I always knew that will happen09:23
karlnapfsonne|work, you arent a liberal person! ;)09:23
blackburndouble gay marriage hmm too liberal for any EU country I'd say09:24
blackburn Log Message:  -----------  move next release date to ~september'1209:24
blackburnsonne|work: hah09:24
blackburnI hope september would be enough09:24
sonne|workwhy not?09:26
blackburnsonne|work: yes would be great release09:28
blackburnsonne|work: however 2.0 is the version I think :D09:28
karlnapfguys, I will be off for some studying, be back later this afternoon/evening, have a good day09:29
blackburnkarlnapf: enjoy your study time ;)09:30
karlnapfthx :)09:31
gsomixblackburn, h09:35
blackburngsomix: hhhhhhh09:35
blackburnsonne|work: it looks like cheng suggests to construct multiple features for group LASSOs09:42
blackburnit contradicts with my approach pretty much09:42
-!- Marty28 [] has joined #shogun09:42
sonne|workblackburn: it is like mkl........09:43
blackburnsonne|work: yeah today everything is like mkl :)09:44
blackburnbut still - what do you prefer?09:44
blackburnto use any structures like task/features relations in machines09:45
sonne|workif you have multiple sets of features it is mkl09:45
sonne|workgroup lasso09:45
sonne|workor whatever name you give it09:45
-!- eric___ [2e1fd566@gateway/web/freenode/ip.] has joined #shogun09:45
eric___hi all09:45
sonne|work(and learn group feature weights...)09:45
blackburnsonne|work: no in group lasso you just use this relations - not learning weight afaik09:46
blackburnsonne|work: actually it is a little bit crazy - you can group outputs, features and tasks09:48
-!- Marty28 [] has quit [Quit: Colloquy for iPad -]09:52
-!- cronor [] has joined #shogun10:01
-!- wiking [] has joined #shogun10:12
-!- wiking [] has quit [Changing host]10:12
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun10:12
gsomixyeah, it's work. I have music10:18
-!- gsomix [~gsomix@] has quit [Remote host closed the connection]10:39
-!- gsomix [~gsomix@] has joined #shogun10:42
-!- av3ngr [av3ngr@nat/redhat/x-arwstdbdsgyjdant] has quit [Quit: That's all folks!]10:49
-!- vikram360 [~vikram360@] has joined #shogun11:01
-!- cronor [] has quit [Quit: cronor]11:53
-!- cronor [] has joined #shogun12:21
-!- cronor_ [] has joined #shogun12:46
-!- cronor [] has quit [Read error: Connection reset by peer]12:47
-!- cronor_ is now known as cronor12:47
-!- karlnapf [] has quit [Ping timeout: 260 seconds]13:34
-!- SeattleDad [322e753f@gateway/web/freenode/ip.] has quit [Quit: Page closed]14:04
-!- karlnapf [] has joined #shogun14:43
* blackburn is getting nervous14:52
eric___wiking: hi again, yesterday you said your paper was accepted ? Which conf ?14:55
-!- pluskid [] has joined #shogun15:45
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]16:17
sonne|workblackburn: why?16:26
sonne|workI guess I know :D16:26
-!- PhilTillet [~Philippe@] has joined #shogun16:27
-!- blackburn [5bdfb203@gateway/web/freenode/ip.] has quit [Quit: Page closed]16:43
eric___what is the syntax for loading svm file for initialize CLibSVMMultiClass ?16:44
eric___CLibSVMMultiClass svm; svm.load(svm_file); give me "No kernel init"16:44
eric___I my trained svm file I have however in the first lines %MultiClassSVM multiclass_type=1; num_classes=10; num_svms=45; kernel='GaussianKernel';16:46
eric___somone here ?16:52
-!- vikram360 [~vikram360@] has quit [Ping timeout: 276 seconds]16:56
-!- vikram360 [~vikram360@] has joined #shogun16:57
eric___allo ?16:58
gsomixeric___, pong16:58
eric___gsomix: :p16:58
eric___gsomix: are you aware of pb reading multiC svm file ?16:58
gsomixeric___, no, hehe. just try to wait for sonney2k or sonne|work16:59
sonne|workeric___: don't use load/save17:04
sonne|workeric___: have a look at the *serialization* examples on how to save things17:05
eric___sonne|work: tx17:11
pluskidhow's the sgvector hell?17:29
-!- uricamic [~uricamic@2001:718:2:1634:f2de:f1ff:fecf:a6a5] has quit [Quit: Leaving.]17:29
pluskidsonne|work: just made a very brief scan of the current SGVector.h, and find two suggestions here17:30
pluskid1. operator=, I think the correct semantic for operator= is first to unref() what current holding and then ref() what is assigned new17:30
pluskid2.  I think clone() method isn't properly implemented: the memory is allocated but no ref-counting is setup, I think the memory allocated here will never be freed, because unref() just returns without freeing the memory when it find m_refcount == NULL17:32
pluskidand the destructor do nothing more than calling unref()17:32
sonne|workpluskid: things are improving :)17:33
sonne|worknot too bad17:33
sonne|workpluskid: why the operator= stuff? I mean we don't need to unref since the object is on stack17:34
pluskidsonne|work: unref maintains the memory (the "vector" member variable), not the object :)17:35
sonne|workclone is not yet done :)17:35
pluskidoperator= means the object already exist17:35
sonne|workand the destructor should just do unref no more17:35
pluskidand you want to assign it to something else17:35
pluskidno, the semantic is this:17:36
sonne|workpluskid: clone is for getting a copy of sgvector (including the memory block)17:36
pluskidSGVector<int> v1 = ...17:36
sonne|workso it is different from copy constructor, assignment operator17:36
pluskidv1 = SGVector<int>(...)17:36
pluskidthen operator= will be called17:36
pluskidso here we need to unref what v1 is already holding, and then ref what is newly assigned17:37
-!- wiking [] has joined #shogun17:37
-!- wiking [] has quit [Changing host]17:37
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun17:37
pluskidsonne|work: about the clone, I'm concerning the newly allocated memory will never be freed, unless you do explicit SG_FREE(v.vector) somewhere17:38
pluskidthe current implementation of clone17:38
sonne|workclone is not yet implemented17:38
pluskidoh, OK17:39
sonne|workclone would be17:39
sonne|workreturn SGVector<T>(clone_vector(vector, vlen), vlen);17:39
sonne|workno more than that17:39
pluskidhmm, true17:40
sonne|workpluskid, regarding what you said above17:40
sonne|workno one thing is copy constructor17:41
sonne|workand the other is assignment operator17:41
sonne|workthings won't be done twice17:41
pluskidcopy constructor is totally different from operator=17:41
pluskidand I think in operator= we need to release previous resource before accepting the new resource17:42
sonne|workwhy that?17:42
sonne|workthe previous thing will still remain17:42
sonne|workif you have17:42
sonne|workSGVector x;17:42
sonne|workSGVector y;17:43
sonne|workand you do x=y17:43
sonne|workx will be killed17:43
sonne|workand y and x will hold the thing17:43
sonne|workso only when y & x are destroyed should refcount be equal to zero17:44
pluskidno, in our current implementation: x and y will hold the thing but the previous content of x will lost17:44
sonne|workahh now I understand17:44
sonne|workso I should just call unref() in operator= as first line and all good right?17:45
-!- wiking_ [] has joined #shogun17:45
-!- wiking_ [] has quit [Changing host]17:45
-!- wiking_ [~wiking@huwico/staff/wiking] has joined #shogun17:45
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 260 seconds]17:45
-!- wiking_ is now known as wiking17:45
pluskidyes, I think so17:45
CIA-113shogun: Soeren Sonnenburg master * r4aafbd1 / src/shogun/lib/SGVector.h : fix assignment operator of SGVector and its clone method -
pluskidhave to sleep now, hope we fix all errors with migrating SGVector soon! :D17:46
pluskidgood night17:46
-!- pluskid [] has quit [Quit: Leaving]17:47
-!- eric___ [2e1fd566@gateway/web/freenode/ip.] has quit [Quit: Page closed]18:19
-!- sonne|work [~sonnenbu@] has left #shogun []18:30
-!- n4nd0 [02893bbe@gateway/web/freenode/ip.] has joined #shogun19:05
-!- karlnapf [] has left #shogun []19:28
-!- PhilTillet [~Philippe@] has quit [Ping timeout: 255 seconds]21:18
-!- blackburn [~qdrgsm@] has joined #shogun21:25
blackburnoh I just noticed version is 2.0, sonney2k21:42
blackburnokay let me try to fix something21:43
-!- cronor [] has quit [Read error: Operation timed out]21:43
blackburnsonney2k: I didn't noticed before that you changed next version21:51
@sonney2kwe are changing too much to no justify a 2.0 release ...21:51
blackburnsonney2k: I agree totally21:52
-!- PhilTillet [] has joined #shogun21:53
blackburnn4nd0: hey21:53
n4nd0blackburn: hey there!21:54
blackburnn4nd0: how are you?21:54
blackburnsonney2k: do you feel yourself as linus when you do merges all day long?21:55
blackburnI do ;)21:55
CIA-113shogun: Chiyuan Zhang master * r43208fe / src/interfaces/modular/Classifier.i : Added python_modular interface for ECOC Random Coding. -
CIA-113shogun: Chiyuan Zhang master * r64efd80 / (2 files in 2 dirs): Added ECOC Random modular interface and a simple python example. -
CIA-113shogun: Sergey Lisitsyn master * r25427d4 / (3 files in 2 dirs): Merge branch 'multiclass-ecoc' of git:// -
CIA-113shogun: Evgeniy Andreev master * r5427a46 / (6 files): removed load/save methods -
CIA-113shogun: Sergey Lisitsyn master * re0cb338 / (6 files): Merge branch 'save_load_Machines' of git:// -
n4nd0blackburn: I am fine, what about you?21:57
blackburnn4nd0: oh I've been trying to leave job whole week :D21:58
n4nd0blackburn: oh, is it difficult?21:58
blackburnn4nd0: yes because of my instability :D21:58
blackburnthey suggested to think about long break (3 months) and I promised to think about it21:59
blackburntwo times damn21:59
n4nd0I see22:00
@sonney2kblackburn, is on the 'wanted' list :D22:00
@sonney2kn4nd0, how is our exam preparation going?22:00
blackburnsonney2k: that OSS shit is so difficult to grasp they want to keep every man that understand anything in it22:01
blackburnI wouldn't say I'm useful there22:01
blackburnhowever I will not work this summer for sure22:02
blackburnI should checked compilation22:02
blackburnsonney2k: could you please open ui/GUIClassifier.cpp:840?22:03
blackburnwhat should be there instead of load?22:03
-!- n4nd0 [02893bbe@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]22:04
-!- hsyal [~harshit.s@] has joined #shogun22:05
blackburnascii I guess?22:05
-!- hsyal [~harshit.s@] has quit [Client Quit]22:06
-!- n4nd0 [02893bbe@gateway/web/freenode/ip.] has joined #shogun22:06
n4nd0sorry guys, my internet sucks so bad ...22:06
@sonney2kn4nd0, too bad yo miss all the jokes we make about you :P22:06
n4nd0sonney2k: the preparation for the exam is going somewhat good, thanks!22:06
n4nd0haha I am lucky that there are logs!22:07
@sonney2kblackburn, we don't have any other choice22:07
@sonney2kI have to get inside...22:07
@sonney2kit is getting cold22:07
@sonney2kbut today is full moon here22:07
blackburnn4nd0: oh I shouldn't tell anyone how did I parse 'sucks so bad'22:07
blackburnsonney2k: is it 'yes'?22:08
@sonney2klooks pretty cool :)22:08
blackburnok I'll patch22:08
n4nd0it is rainining all the days in Spain since I came ... I cannot believe this :(22:08
shogun-buildbotbuild #820 of libshogun is complete: Failure [failed compile]  Build details are at  blamelist:, gsomix@gmail.com22:09
blackburnshogun-buildbot: do not blame me bastard22:09
n4nd0sonney2k: I checked before the work that Nico has started22:10
n4nd0I like it :)22:10
blackburnn4nd0: what is it?22:10
blackburnn4nd0: have you ever seen 'explicit' keyword in C++?22:11
n4nd0blackburn: a short document with some notes about SO and some code in gist22:11
n4nd0blackburn: I think I have yes22:11
blackburnn4nd0: wow that should help you22:12
n4nd0I'm looking forward to start devoting my time to the project!!22:12
-!- PhilTillet [] has quit [Ping timeout: 245 seconds]22:12
blackburnn4nd0: 2 days I guess?22:13
-!- PhilTillet [] has joined #shogun22:13
n4nd0more or less22:14
n4nd02 days for the exam (not counting today)22:14
n4nd0Tuesday back to Sweden, and voila!22:15
@sonney2kn4nd0, will you stay in sweden afterwards?22:16
n4nd0yes :D22:16
n4nd0it's my home22:16
@sonney2kn4nd0, so you are connected again :)22:16
blackburnn4nd0: are you going to pursuit phd?22:16
n4nd0haha yes22:17
@sonney2kbut actually w/o internet it is good to focus :)22:17
n4nd0blackburn: sure22:17
blackburnyeah sonney2k22:17
blackburnreally good to focus22:17
* sonney2k is being distracted by something called chat22:17
blackburnI'll visit dacha this summer to focus :D22:17
n4nd0blackburn: hopefully I'll be able to start by Fall 201322:17
n4nd0blackburn: and you?22:17
blackburnn4nd0: I am thinking about foreign phd..22:18
@sonney2kn4nd0, as long as you use shogun you will succeed :D22:18
blackburnFall 2014 for me22:18
@sonney2kyou know it has been written by the brightest minds of this century :D22:18
n4nd0sonney2k: once we have HM-SVM I want to try for a game we did it in a course!22:19
n4nd0I wonder whether it would work or not22:19
@sonney2kthere are many nice applications...22:19
CIA-113shogun: Sergey Lisitsyn master * rc688796 / src/shogun/ui/GUIClassifier.cpp : Hopefully fixes static serialization -
blackburnsonney2k: my chef said it is ok I do opensource - 'who knows you may become a millionaire because of it' :D22:21
blackburnsounds like a plan isn't it?22:21
@sonney2kyou mean like bill gates ::D22:22
blackburnsonney2k: things have actually changed and good small idea can impact too I think22:23
blackburnn4nd0: sonney2k: can you believe 20-30% of people here participate in a financial pyramid scheme?22:24
* n4nd0 is reading pyramid scheme's article in wikipedia22:25
shogun-buildbotbuild #821 of libshogun is complete: Success [build successful]  Build details are at
blackburnn4nd0: like you involve people that pay you to be in a pyramid22:26
n4nd0it sounds like kind of fraud22:26
n4nd0am I wrong?22:26
blackburnand they involve others22:26
blackburnso money is going from the bottom throught to the top :D22:26
-!- PhilTillet [] has quit [Ping timeout: 240 seconds]22:26
blackburnn4nd0: yes it is illegal in *NORMAL* countries22:26
n4nd0you should come to Sweden22:27
blackburnn4nd0: should be kind of normal country right? ;)22:30
n4nd0I think it is a very good country22:30
n4nd0the only con is the lack of light during winter22:31
n4nd0too bad22:31
blackburnn4nd0: hmm I am curious what are opportunities that would fit to me?22:31
blackburndo they provide some stipend for phds?22:31
blackburnI am not a millionaire yet  :D22:32
n4nd0blackburn: stipends? no no22:32
n4nd0in Sweden it is like a job22:32
blackburnn4nd0: what is salary?22:32
blackburnaverage ;)22:32
-!- PhilTillet [] has joined #shogun22:40
-!- cronor [] has joined #shogun22:56
blackburnsee you tomorrow (expect a lot commits from me ;)23:23
-!- blackburn [~qdrgsm@] has quit [Quit: Leaving.]23:23
-!- PhilTillet [] has quit [Ping timeout: 240 seconds]23:30
n4nd0good night people23:49
--- Log closed Sat May 05 00:00:37 2012