Open in new window / Try shogun cloud
--- Log opened Sun Feb 26 00:00:19 2012
-!- ishaanmlhtr [~chatzilla@] has quit [Ping timeout: 260 seconds]00:26
-!- n4nd0 [] has quit [Quit: leaving]01:52
-!- ishaanmlhtr [~chatzilla@] has joined #shogun05:49
-!- ishaanmlhtr [~chatzilla@] has quit [Quit: ChatZilla 0.9.88 [Firefox 9.0.1/20111220165912]]07:33
-!- delcoder [~ishaan@] has joined #shogun09:43
-!- blackburn [~qdrgsm@] has joined #shogun10:09
-!- blackburn [~qdrgsm@] has quit [Quit: Leaving.]12:06
-!- wangyi160 [3ac2e06c@gateway/web/freenode/ip.] has joined #shogun12:20
wangyi160anybody here12:21
wangyi160I have a question, I am using java interface, where is Load.load_numbers, it is very confusing in the example12:22
wangyi160I use DoubleMatrix.loadAsciiFile to replace that, but seems wrong12:22
-!- blackburn [~qdrgsm@] has joined #shogun12:26
-!- wangyi160 [3ac2e06c@gateway/web/freenode/ip.] has quit [Quit: Page closed]12:32
-!- n4nd0 [] has joined #shogun12:34
n4nd0blackburn: hi!12:36
n4nd0blackburn: I cannot manage to fix a compile error I am getting for the example with the Mahalanobis distance12:37
blackburnheh let me try to help12:37
n4nd0blackburn: the include directive to the header file MahalanobisDistance.h is working fine12:37
n4nd0blackburn: but I still get 'CMahalanobisDistance' was not declared in this scope12:38
n4nd0blackburn: and the class is defined in that header12:38
blackburnplease paste source somewhere12:38
n4nd0and the header goes here12:40
n4nd0I thought it might be cause by the ifdef HAVE_LAPACK directive12:41
n4nd0put the error persisted when I commented it out :S12:41
blackburnok let me think12:43
blackburnn4nd0: does your snippet work with euclidean distance?12:44
blackburnstrange, can't see any error12:46
blackburncould you please paste the .cpp as well?12:46
n4nd0here it is
n4nd0blackburn: I am doubting whether the ifdef HAVE_LAPACK in the .cpp file might be a problem12:53
blackburnn4nd0: shouldn't be if you have lapack12:53
blackburnit should be here for sure12:53
n4nd0blackburn: ok12:54
blackburnn4nd0: chekc interfaces/modular/Distance_includes.i12:57
blackburnand interfaces/modular/Distance.i12:57
blackburnand add your distance to make it available from modular (python,etc)12:57
n4nd0blackburn: I have done it now but the error is still there13:02
n4nd0but I guess I have to compile shogun first again13:02
blackburnno, it won't help with this error13:02
n4nd0mmm then it's bad13:03
blackburnI currently can't see any reason why it fails :(13:03
n4nd0blackburn: you were right, it didn't make the trick to compile :(13:20
n4nd0blackburn: I have no idea why can it be wrong13:20
blackburntry to use it from python13:20
blackburnwhen things could get clearer13:20
n4nd0blackburn: you like more python definetely ;P13:25
blackburnsure, but I just want to check your distance is being included13:25
blackburnbtw you will like it too if you don't13:26
n4nd0I like it but I have more experience with C++ now13:26
n4nd0but I want to improve my python skills13:27
n4nd0I'll be back later13:28
blackburnI'll try to find out error13:32
-!- wangyi160 [3ac2e06c@gateway/web/freenode/ip.] has joined #shogun15:15
blackburnwangyi160: hi15:22
wangyi160Oh, finally, I found it in the examples/java_modular folder... it seems I should import the whole folder into eclipse..15:46
wangyi160why no one answer me... no help, only help myself15:47
-!- wangyi160 [3ac2e06c@gateway/web/freenode/ip.] has quit [Quit: Page closed]15:47
* blackburn is curious where is the question :D16:12
n4nd0blackburn: so the instruction from shogun.Distance import MahalanobisDistance works fine from python16:29
n4nd0blackburn: that is at least a good new I guess :)16:29
blackburnyeah good news16:33
n4nd0blackburn:time to debug then! it looks like my Mahalanobis distance gives always zero as a result :S16:41
blackburnhaha good distance16:42
n4nd0haha eveything is close :P16:42
n4nd0blackburn: now that I think of it, is there any way to debug the code written in C++ from a example written in python?16:43
n4nd0blackburn: I don't know if I make my point clear...16:43
blackburnwell I usually do some debug print16:43
blackburnit usually helps, but sometimes C++ example is required..16:44
n4nd0blackburn: all right, so just using printf/cout in the C++ code I will get the traces when I execute the example with python?16:44
blackburnno, use SG_PRINT16:44
n4nd0SG_PRINT("Test message", stdout)?16:47
blackburnw/o stdout part16:47
n4nd0does it work with printf formatting style?16:47
n4nd0SG_PRINT("%d", 5)?16:47
n4nd0cool, thank you16:47
-!- delcoder [~ishaan@] has quit [Ping timeout: 252 seconds]17:10
n4nd0blackburn: another question :P17:12
n4nd0blackburn: so when one is working with just one file within the project doing some debugging, like I am doing right now with this mahalanobis distance17:13
n4nd0blackburn: is there any way to just build that part so the modifications can be tested faster?17:13
blackburnnot really, when you change .h interfaces will recompile17:14
blackburnbut changing .cpp's will not cause interfaces recompile - would be much faster17:14
n4nd0but it is the same process then17:15
n4nd0make, make install?17:15
blackburnjust try to change only cpp17:15
blackburninterfaces won't recompile17:15
blackburnit is the slowest step here17:15
n4nd0to run configure also makes the recompilation right?17:16
n4nd0I mean that it forces it17:16
-!- delcoder [~ishaan@] has joined #shogun17:32
n4nd0blackburn: man, why can I do RealFeatures(fm) if fm contains just one feature vector?17:33
n4nd0ups can't I wanted to say17:34
blackburnit should be matrix17:34
blackburnI guess numpy treated it as vector17:34
n4nd0I will try to make a cast then ... but I am actually loading it using LoadMatrix17:36
blackburnnumpy tends to do this thing sometimes17:36
-!- blackburn [~qdrgsm@] has quit [Quit: Leaving.]17:46
-!- blackburn [~qdrgsm@] has joined #shogun18:39
n4nd0blackburn: I was checking the copy constructor of the class SGVector19:09
blackburnwhat's up with it?19:09
n4nd0blackburn: shouldn't it copy the contents to another portion of memory?19:09
n4nd0blackburn: I don't think it does that19:10
blackburnI guess clone is supposed to do that19:10
n4nd0yeah I tried with that too19:10
n4nd0but clone is actually cloning the T*, not the SGVector structure itself19:11
blackburnwhat do you mean?19:12
blackburnI can see it does exactly what you want..19:12
n4nd0you mean clone_vector in SGVector?19:13
n4nd0maybe there is a method clone I didn't see19:13
blackburnyes, clone19:13
blackburnnot clone_vector19:14
n4nd0aha, all right, I didn't see that one19:14
blackburnno problem :)19:17
-!- n4nd0 [] has quit [Quit: leaving]19:18
-!- n4nd0 [] has joined #shogun19:18
-!- n4nd0 [] has quit [Client Quit]19:21
-!- delcoder [~ishaan@] has quit [Quit: Leaving]19:36
-!- n4nd0 [] has joined #shogun20:10
n4nd0blackburn: fixed :) I am obtaining nice distance values now20:51
n4nd0one question about the design20:52
n4nd0this Mahalanobis distance looks to be defined both for distance from a point to a distribution and for distance between two distributions20:53
n4nd0is it interesting to have both or point to distribution is ok?20:53
blackburnI think it should be distance rhs -> lhs pointwise20:54
blackburni.e. lhs should be considered as distribution20:55
n4nd0the function compute receives two indexes as arguments because of inheritance, one of them is not used then20:56
n4nd0no problem with that right?20:56
blackburnit is ok I guess20:58
n4nd0I found it weird that I can use the code with no problem from python but it was not working before in the C++ example21:00
blackburndon't worry we will find out what is wrong with it21:01
n4nd0blackburn: I will do a pull request then22:00
n4nd0blackburn: by the way, should I write sth in author and copyright?22:00
blackburnyes, your name :)22:00
blackburnI hope it fits into one line ;)22:01
blackburnjoke :)22:01
n4nd0haha it does ;)22:01
n4nd0a bit longer than the usual ones, but it does22:01
blackburnwhat is J. btw?22:02
blackburnbtw my name has patronymic, so I am Sergey Olegovitch Lisitsyn :) going long as well22:03
blackburnI see22:06
n4nd0Fernando Jo?e Iglesias Garc?a22:06
blackburnI heard brazil guys have even more names22:06
n4nd0yeah it might be22:07
blackburnI am curious22:07
blackburndo you understand portuguese?22:08
n4nd0I can understand quite a lot actually22:08
blackburnI see22:08
n4nd0but I used to hang out with people from Brazil when I lived in Spain22:08
blackburnI used to try to learn portuguese22:08
n4nd0oh cool22:08
blackburnand no time anymore to do such things haha22:09
n4nd0do you manage to understand then "ai se eu te pego"?22:09
blackburnno, do not understand anything22:09
blackburnwell soeren has been learning russian for X years22:10
blackburnand I have been learning german at school22:10
n4nd0and do you manage with German and he with Russian?22:10
blackburnand actually we both do not really understand each foreign lang :)22:10
n4nd0I think Russian must be really difficult!22:11
blackburnI'll took some spanish lessons from you ;)22:11
blackburnno idea, could be :)22:11
n4nd0I am still fighting against Swedish :O22:11
blackburnhow long have you been in Sweden?22:12
n4nd0this is my second year studying here22:12
blackburnI see22:12
n4nd0I have been for about 18 months now I think22:12
blackburnah not really long22:12
n4nd0I came back to Spain during summer and so on22:12
blackburnso you are going back to spain this summer again?22:13
blackburnI wonder why you have choosed to study in sweden :)22:13
n4nd0the university is really well-known in the university I came from22:14
blackburnI see22:15
n4nd0I wanted to go to the nordic countries22:15
n4nd0but I was not very sure to which one22:15
blackburnnot cold enough? :)22:15
n4nd0so this university made the final decision22:15
blackburnwe have some cold towns here22:15
blackburnI live in not-very-cold-enough one though22:16
n4nd0it is fucking cold in winter but the worst thing for me is the sun ...22:16
blackburnwhat's up with the sun?22:16
n4nd0one needs almost to pray to see it during winter22:16
n4nd0too few hours of sun for me22:17
n4nd0never thought that it would affect me but it actually does22:17
blackburnfunny, never thought it can affect so hard22:18
n4nd0where do you come from in Russia? I have a couple of friends from Moscow and Volgograd living here in Stockhollm22:21
blackburnI live in Togliatti22:25
blackburnand study in Samara22:25
blackburnso going back and forth22:25
blackburnit is pretty near :)22:25
blackburnI'm going to commit some things too soon22:34
n4nd0what have you been working on?22:36
blackburngeneric multiclass things22:37
n4nd0haven't heard about them22:38
n4nd0what are they in particular?22:38
blackburnah just some classes making possible to use binary classifiers22:39
blackburnas multiclas22:39
blackburnOvR, OvO22:39
n4nd0time to bed22:53
blackburngood night22:53
n4nd0good night22:53
n4nd0and thank you for all your help ;)22:53
-!- n4nd0 [] has quit [Quit: leaving]22:54
blackburnyou are welcome22:56
CIA-64shogun: Sergey Lisitsyn master * r1ddb633 / (10 files in 2 dirs): Introduced multiclass generics -
shogun-buildbotbuild #557 of libshogun is complete: Failure [failed compile]  Build details are at  blamelist: blackburn91@gmail.com23:20
CIA-64shogun: Sergey Lisitsyn master * rdb9a699 / (2 files): Hotfix for MC machines -
shogun-buildbotbuild #558 of libshogun is complete: Success [build successful]  Build details are at
--- Log closed Mon Feb 27 00:00:19 2012