Open in new window / Try shogun cloud
--- Log opened Wed Feb 29 00:00:19 2012
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun00:04
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]00:30
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]00:48
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun05:39
-!- sonne|work [~sonnenbu@194.78.35.195] has quit [Ping timeout: 240 seconds]06:55
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun07:11
-!- blackburn [~qdrgsm@188.168.2.216] has joined #shogun07:55
-!- blackburn [~qdrgsm@188.168.2.216] has quit [Ping timeout: 252 seconds]08:12
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun08:23
n4nd0hi there08:37
n4nd0is there any nice function already implemented in shogun to concatenate matrices in C++?08:38
sonne|workno08:49
sonne|workone should rather try to avoid this for large feature matrices...08:50
sonne|worki.e. for computing cov you should rather add a new cov function that operates on two matrices :)08:50
n4nd0ok, I see08:55
n4nd0I will first implement it in the class MahalanobisDistance and if it is better later I can move to another where it is more appropriate08:56
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]08:58
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 265 seconds]09:10
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun09:45
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun10:01
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]10:42
sonne|workn4nd0: it would make sense to just put it next to the get_cov from dotfeatures :)10:49
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 276 seconds]11:01
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun11:02
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun12:20
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun13:34
blackburnsonne|work: hey, howdy?14:01
blackburnn4nd0: do you need any other suggestions what to work on? ;)14:01
sonne|workhowdy cowboy ;-)(14:02
blackburnsonne|work: will you store current proposal at shogun-publicity?14:10
sonne|workwhen I made some more progress yes14:10
blackburnis it raw still?14:11
sonne|workbtw, we have some new great ideas from vojtech :)14:11
sonne|workyes14:11
blackburngreat ideas from vojtech? I am curious14:11
blackburnso what are the ideas?14:11
sonne|workI will commit this in the evening14:11
blackburnwill he be able to mentor 2 students?14:12
blackburnsonne|work: sagen sie mal ideas!14:12
blackburn:D14:12
blackburnyes I know it is not a proper deutsch ;)14:13
sonne|workwell we don't know if we get good proposals for all ideas...14:13
blackburnsonne|work: away     : ? db & Biotek - SOD ? -- hey now I know what is the music you are listening to! :D14:16
blackburntell me the ideas!14:17
blackburndamn once upon a time I will get your phone number and will call you to make you answer ;)14:19
n4nd0blackburn: sure! tell me14:56
n4nd0blackburn: I am a bit stuck with some university work but I hope I will be back soon to fix the Mahalanobis stuff and after I can take a look to another thing14:57
blackburnn4nd0: I'd have to think about it though :)14:58
blackburnsame thing here with work and university14:58
n4nd0blackburn: ah ok! I thought you already came up with something, I took a quick look yesterday to GPs in scikit-learn15:00
blackburnnot really but I had some ideas before and 'll try to remind it during the spare time15:01
n4nd0cool15:02
n4nd0what do you study in the university btw?15:02
blackburnwell some shit I do not really like to study hah15:02
blackburnofficially I have major in mathematics and computer science15:03
blackburnhard to translate these parts of russian education to your standarts :)15:04
blackburnbut I guess I can say it is major15:04
n4nd0haha I know what you mean with standards15:05
n4nd0my Spanish studies are not really within any normal standards either ...15:05
blackburnif you have talked to some russians you probably know we can't choose what to study :)15:06
n4nd0oh really?15:06
n4nd0didn't know about that! I have met some Russians around here but never heard it15:06
n4nd0that doesn't sound good15:07
blackburnso I just have to study what they say to study :)15:09
blackburncurrently it is some optics (again) and DSP15:09
blackburnand also some software engineering stuff15:09
n4nd0I am taking right now a course from Spain in software engineering too15:10
n4nd0personally I don't like it that much, too theoretical from my point of view15:10
blackburnmy SE course here is a useless crap15:11
blackburnI have never seen such stuff they treat as software engineering 'ways' anywhere :)15:12
n4nd0:)15:13
blackburnn4nd0: funny thing I have no machine learning courses here as well15:16
blackburnand I did some unofficial course by myself last year :D15:17
n4nd0blackburn: it's not nice that they don't have ml courses15:18
blackburnhttps://dl-web.dropbox.com/get/Public/ml/classification-NB%2CID3.pdf15:18
n4nd0did you take those courses in the university or somewhere else?15:18
blackburnI bet you won't understand much ;)15:18
blackburnin russian hah15:18
n4nd0haha15:19
blackburnwell I just asked my fellows if they want to study it a little15:19
n4nd0the link doesn't wotk either15:19
blackburnreally?15:19
blackburnand did 5 lectures on some basic ML15:19
n4nd0cool15:19
blackburnhttp://dl.dropbox.com/u/10139213/ml/classification-NB%2CID3.pdf15:19
blackburnhm sorry yes15:19
blackburnright one ^15:19
n4nd0I like slides with latex :)15:20
blackburneveryone do :)15:20
n4nd0I got to know ml last year when I came to Stockholm15:21
n4nd0never saw something similar before in my courses in Spain15:21
n4nd0well maybe something similar to classification with detectors in digital communications15:21
blackburnhardest thing here was to fit python classifiers in one slide :D15:21
n4nd0haha15:22
blackburnyeah ML is not known everywhere yet15:22
n4nd0but the funniest thing is that some people in my home university makes some reasearch on svms15:22
n4nd0so it is like ... wtf you should let us know about that stuff!15:22
blackburnhaha yeah15:23
n4nd0anyway15:23
blackburnpeople here in my one do not know svms at all15:23
n4nd0they have some pretty cool courses at KTH on machine learning15:23
n4nd0there is a machine learning master actually15:23
blackburnthey prefer to do something like 'solving neyman-pearson task with fisher discriminant analysis'15:24
blackburnin other words to do stuff everyone did in 1960s15:24
blackburn:D15:24
n4nd0yeah, I know what you mean15:24
blackburnthey think they can really control probabilities with these outdated stuff :)15:25
blackburnn4nd0: there is a soviet legacy everywhere15:27
blackburnwe call it 'sowok' :)15:27
n4nd0does it stand for sth?15:28
blackburnsoviet union = sowetskiy soyuz -> sowok15:29
n4nd0haha ok15:30
blackburnthis word also stands for scoop for dust :)15:30
blackburnoh it seems I don't want to work at all15:32
n4nd0haha nice coincidence the one for the dust15:32
n4nd0same here, I am scared of a problem I have to hand in :S15:32
blackburnlets check whether my manager could notice I have nothing to work on15:32
blackburnn4nd0: what is the problem?15:36
n4nd0blackburn: wireless communications stuff15:37
blackburnI see15:37
n4nd0I actually enjoy the contents of the course, they are interesting15:38
wikingblackburn: the whole pdf is in russian! :)15:46
wikingdo something about it :)15:46
blackburnwiking: well I have presented it to russian audience ;)15:47
wiking:>15:47
blackburnactually my english is pretty bad, but others have even worse15:48
blackburnso it would be way too funny15:48
n4nd0c'mon, you write quite well around here, your English cannot be bad!15:49
wikingblackburn: you've gotta practice mate :)15:50
wikingah fuck yeah i should resend the email to alex15:51
wikingwhat's sonney2k 's email?15:51
wikingsonne@debian.org uses this?15:52
blackburnyes15:52
wikingcool15:52
wikingok done \o/15:53
wikingbbl15:59
blackburnnice, I hope he will answer this time15:59
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]15:59
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun16:49
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed]16:57
-!- puneet [~chatzilla@101.63.76.203] has joined #shogun17:10
-!- ishaanmlhtr [~chatzilla@115.241.250.55] has joined #shogun17:13
-!- ishaanmlhtr [~chatzilla@115.241.250.55] has quit [Quit: ChatZilla 0.9.88 [Firefox 9.0.1/20111220165912]]17:19
-!- puneet [~chatzilla@101.63.76.203] has quit [Ping timeout: 244 seconds]17:23
-!- delcoder [~ishaan@115.241.158.121] has joined #shogun17:24
-!- delcoder [~ishaan@115.241.158.121] has quit [Ping timeout: 260 seconds]17:34
n4nd0it looks like I finally convinced a friend of mine to use shogun to help with the work on his thesis :)17:44
-!- bais [~bais@s213-103-215-230.cust.tele2.se] has joined #shogun19:00
baishi!19:02
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking]19:06
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun19:06
n4nd0bais: hey!19:08
baisi was considering to use shogun to test my own implementation of a string kernel.19:09
baiswould it be particularly hard to plug in my own kernel function in a SVM using shogun?19:10
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 245 seconds]19:11
n4nd0bais: I am not completely sure about it, I guess it depends on the complexity of your kernel19:14
n4nd0but it shouldn't be a big deal at first sight19:14
n4nd0take a look at this one, it is particularly short19:14
n4nd0http://www.shogun-toolbox.org/doc/en/current/FixedDegreeStringKernel_8cpp_source.html19:14
n4nd0but I guess it would be more interesting if you could use your own implementation in python/octave so you can try it faster19:15
n4nd0but I don't know if that can be done without implementing it in C++ :(19:15
baisoh i see.. my intention was exactly to use an octave implementation, so speed up the development and test the results without having too much trouble in the coding process19:18
baisat first at least..19:20
* n4nd0 19:30
* bais thanks n4nd0 for his help, and thinks n4nd0 is a valid member of the shogun community19:30
n4nd0bais: haha I have just been around here for a couple of weeks19:35
n4nd0member sounds more important than that :P19:35
baisi will continue reading some documentation and come back in case i have other problems / questions!19:41
baistnx bye!19:41
-!- bais [~bais@s213-103-215-230.cust.tele2.se] has left #shogun []19:41
-!- Ram108 [~amma@14.99.88.181] has joined #shogun20:18
-!- Ram108 [~amma@14.99.88.181] has quit [Client Quit]20:18
n4nd0sonne|work: for the covariance stuff, in order to decide whether rhs and lhs are equal, is it enough with lhs != rhs20:45
n4nd0sonne|work: or should I loop and compare element by element?20:45
n4nd0sounds like sort of stupid question :(20:46
@sonney2kn4nd0, yes20:58
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]23:05
--- Log closed Thu Mar 01 00:00:19 2012