Open in new window / Try shogun cloud
--- Log opened Tue Jan 31 00:00:19 2012
CIA-11shogun: Sergey Lisitsyn master * re92461c / src/shogun/converter/KernelLocallyLinearEmbedding.cpp : Removed junk -
CIA-11shogun: Sergey Lisitsyn master * rff2eba8 / applications/edrt/ : unParallel swissroll example -
CIA-11shogun: Sergey Lisitsyn master * rc36dfbc / (11 files in 4 dirs): PTHREAD ifdefing -
-!- blackburn [~qdrgsm@] has quit [Quit: Leaving.]01:41
-!- dfrx [] has joined #shogun05:36
-!- dfrx [] has quit [Quit: Leaving.]05:39
-!- dfrx [] has joined #shogun05:46
-!- dfrx [] has left #shogun []06:16
-!- dfrx [] has joined #shogun06:16
-!- dfrx [] has left #shogun []06:16
-!- dfrx [] has joined #shogun06:16
-!- dfrx [] has left #shogun []07:00
-!- dfrx [] has joined #shogun07:00
-!- dfrx [] has left #shogun []07:01
-!- dfrx [] has joined #shogun07:01
-!- dfrx [] has left #shogun []07:01
-!- dfrx [] has joined #shogun07:31
-!- dfrx [] has left #shogun []07:37
-!- dfrx [] has joined #shogun07:49
-!- dfrx [] has quit [Quit: Leaving.]11:41
-!- Ram108 [~amma@] has joined #shogun14:54
CIA-11shogun: Soeren Sonnenburg master * r870172a / (14 files in 5 dirs):15:20
CIA-11shogun: Merge pull request #360 from karlnapf/master15:20
CIA-11shogun: introduced standard cross-validation splitting (+6 more commits...) -
-!- Ram108 [~amma@] has quit [Ping timeout: 252 seconds]16:35
-!- Ram108 [~amma@] has joined #shogun16:59
-!- blackburn [~qdrgsm@] has joined #shogun17:09
-!- blackburn1 [~qdrgsm@] has joined #shogun17:56
-!- blackburn [~qdrgsm@] has quit [Ping timeout: 255 seconds]17:58
-!- blackburn1 [~qdrgsm@] has quit [Ping timeout: 276 seconds]18:37
-!- blackburn [~qdrgsm@] has joined #shogun18:39
-!- blackburn1 [~qdrgsm@] has joined #shogun19:10
-!- blackburn [~qdrgsm@] has quit [Ping timeout: 240 seconds]19:14
blackburn1sonney2k: around?19:24
-!- blackburn1 is now known as blackburn19:24
blackburnI guess I have to place snapshot on the wbsite for now?19:26
-!- Ram108 [~amma@] has quit [Ping timeout: 245 seconds]19:36
-!- Ram108 [~amma@] has joined #shogun19:56
-!- cwidmer [8ca3fe9d@gateway/web/freenode/ip.] has joined #shogun20:13
-!- sonne|work [~sonnenbu@] has quit [Ping timeout: 276 seconds]20:14
@sonney2kblackburn, done20:18
blackburnsonney2k: done with snapshot?20:18
@sonney2kbtw we have 2 new potential mentors20:18
@sonney2kone willing to mentor GP integration20:18
@sonney2kolivier stegle20:19
blackburngaussian processes?20:19
@sonney2kand one for doing structured output framework20:19
@sonney2knico goernitz20:19
@sonney2kI don't think we should have more20:19
blackburnhmm he is from tuebingen too?20:19
@sonney2kolivier is from tuebingen20:19
blackburnheh full of germans20:19
-!- Ram108 [~amma@] has quit [Quit: Ex-Chat]20:24
-!- sonne|work [~sonnenbu@] has joined #shogun20:29
@sonney2kblackburn, you should make the description a list20:35
blackburnsonney2k: does it support html?20:36
@sonney2kand add pca / kpca20:36
@sonney2kno but markdown syntax20:36
blackburnsonney2k: how to describe unordered list with this syntax?20:36
@sonney2kgoogle :)20:37
@sonney2kbut IIRC20:37
@sonney2k- a20:37
@sonney2k- b20:37
blackburnsonney2k: okay thanks20:37
blackburnsonney2k: we got some doubts about our abstract20:37
blackburnsonney2k: could you please judge it? :)20:37
blackburnthere is a double SHOGUN referring20:38
@sonney2kblackburn, btw double affiiliations won't work with jmlr20:39
@sonney2ktell that to chris20:39
@sonney2khe can only use one20:39
blackburncwidmer: ^20:39
@sonney2kand the other as footnote20:39
@sonney2kblackburn, regarding abstract don't use SHOGUn twice20:40
@sonney2kjust write with bindings20:40
blackburnWe combine speed with ease of use, as our implementations are in C++, while providing bindings to other common languages such as Python, MATLAB, Octave, Java, Ruby, Lua, C\#.20:40
blackburnwith bindings20:40
blackburnthe problem is we can't say we wrote bindings :)20:41
-!- f-x [~user@] has joined #shogun20:43
@sonney2kblackburn, you don't say that you wrote them - you only say these are provided20:50
blackburnso how to say it better20:50
blackburnwith provided bndings?20:50
@sonney2kgood as is20:50
@sonney2kno porblom20:51
@sonney2kscnr :D20:51
f-xblackburn, sonney2k, you guys here? hi everybody!20:54
blackburnf-x: hi, sure, we were talking :)20:55
f-xyeah hello there blackburn! :) been a while actually20:55
@sonney2kso f-x wassup?20:56
f-xhi sonney2k, been going through the exciting 1.1.0 over the last couple of days20:57
f-xnot gone too deep yet, just doing it a bit slow. time management and all required a bit20:57
blackburnjust like old times on #shogun20:59
f-xme being slow? :) yeah, it is!20:59
f-xactually was thinking of some short/medium-term plan which you guys might have20:59
f-xso i could get myself back on track a bit20:59
f-xanything, sonney2k, blackburn?21:00
@sonney2kf-x, well you could continue your vw / shogun interaction21:00
f-xsonney2k: that's the aim. thought i might need some kind of a familarity phase seeing the changes shogun has undergone21:01
blackburnwell we have a lot of ideas..21:01
@sonney2kf-x, none really in the online feature stuff21:01
@sonney2kf-x, so you as the author should easily be able to continue21:02
f-xah alright then.. i'll think of some kind of road map21:02
f-xgood to know, sonney2k21:02
f-xand blackburn you're welcome to suggest ideas in core shogun too if you need some help21:02
@sonney2kf-x, of course if you are looking for easy tasks you could aswell translate a new linear svm solver from octave to c++ :)21:03
blackburnf-x: yes I would formulate some if you need some tasks21:03
f-xsonney2k: sure, anything.. i want to have a brief spell so i know how to organize my time with work/shogun/other activities21:03
blackburnf-x: have you seen we -> to github issues?21:04
f-xgo ahead, blackburn..21:04
f-xand a nope to that question21:04
@sonney2kf-x, the newton svm for example
blackburnyou should check, we have wrote down some ideas (
blackburntoo briefly but we can discuss21:05
@sonney2kf-x, it really depends on what you want to do21:05
@sonney2kwe also need e.g. multiclass machines21:05
@sonney2kor some image related features21:05
@sonney2k(hog, ...)21:05
blackburnsonney2k: hey I do not like integrating hog here21:06
@sonney2kf-x, so if you have an application in mind or some preferences let us know21:06
f-xsonney2k: at this moment, mainly something to get used to the new 'style' of shogun, if it has changed recently21:06
blackburnsonney2k: we would end with doing our own OS :D21:07
@sonney2kf-x, style didn't change21:07
f-xeg. how we moved on to SGVector, SGMatrix etc.. things like that21:07
f-xand memory management, eg.SG_MALLOC21:07
@sonney2kf-x, we converted most of the code now21:07
@sonney2kto use SGVector/Matrix etc21:07
@sonney2kand everything uses SG_MALLOC21:07
@sonney2kbut that's mostly it21:08
f-xnice.. i'll look at some code to get a feel of it.. and probably that SVM should be ok for starters21:08
@sonney2ksome bigger changes requiring newer versions of swig / python that are not really user visible (but faster - hopefully more elegant)21:08
@sonney2kf-x, yeah that would certainly be an easy job21:09
f-xsonney2k, perfect21:09
f-xi'll get started on it tomorrow. good to have some kind of plan now.. thanks!21:09
@sonney2knp :)21:11
@sonney2kI am getting tired already *yawn*21:11
blackburnsonney2k: lets pray for my paper before going sleep :D21:11
f-xour prayers are with you, blackburn21:12
f-xand the issues page is so nice21:12
blackburnf-x: hah :)21:12
blackburnf-x: I'm trying to present my collection of algorithms as toolkit to JMLR mloss21:13
f-xblackburn, seriously amazing stuff21:13
f-xyou have it more or less ready?21:13
blackburnf-x: what is ready?21:14
f-xthe paper?21:14
blackburnah yes21:14
blackburnI hope to submit it today21:14
blackburnI got afraid of the message above21:15
f-xlooked like some weird map in nethack21:15
* sonney2k is going to compose a song for blackburn named "praising the reviewers"21:15
blackburnsonney2k: seriously I've read reviews for shogun mloss paper and it makes me think they would say 'who are you guy? what is that shit?'21:16
f-xi say you'd do just fine, anyway your efforts are really appreciable, blackburn21:17
blackburnhah I hope so21:17
@sonney2kblackburn, you will get a mixture of the above (reviewers)21:17
@sonney2kalright f-x, blackburn - nice meeting you again. time to sleep for me21:18
blackburnsee you21:18
f-xsee you sonney2k21:18
blackburnno idea what do you mean mixture of the above :)21:18
f-xyour opinion and my opinion, i think he meant21:18
f-xmixed reviews :)21:18
blackburnwould be nice to not having mixture of 'who fucking are you?' and 'what is the shit you present?' :D21:19
blackburnf-x: that's a current presubmit draft21:19
f-xhehe i'm sure they won't use those words _exactly_ :)21:19
f-xooh lemme see21:19
blackburnahha yeah sure21:19
f-xwow blackburn, did you meet cwidmer?21:20
blackburnf-x: not really, why I'd have to?21:20
f-xjust curious21:20
f-xnaah, just a question which struck me21:20
blackburnah no21:20
blackburnI've never been abroad actually :D21:20
f-xblackburn, me neither :-/21:21
f-xblackburn: nice page of edrt you have there21:23
blackburnugly :)21:23
f-xlooks like a proper research page :)21:23
blackburnf-x: btw feel free to do any remarks21:24
blackburnlooks pretty raw for me still21:25
f-xblackburn, lack of energy and a dire lack of knowledge mean i can't make any comments regarding the content yet21:25
f-xbut i actually thought the overall 'feel' is quite good21:25
blackburn:) I see, no problem21:25
f-xand btw blackburn, you know why we capitalize SHOGUN? does it stand for anything?21:26
blackburnf-x: Soeren + Gunnar21:26
blackburnno idea why SHO21:27
blackburnsounds better :)21:27
f-xShoeren :)21:27
f-xgood performance too, i see (edrt)21:28
blackburnwell should be best21:29
blackburnbut sklearn is too fast on toy data21:29
blackburnand matlab is going fast too on high-dim data21:29
blackburnwell it will be faster later21:29
f-xbut its terrible in isomap (sklearn)21:30
blackburnnot really terrible21:30
f-x2-3 times slower21:30
f-xoh i have to sleep now.. but looking good so far blackburn. all the best with it21:37
blackburnf-x: thanks! sleep well21:37
f-xsee you21:38
-!- f-x [~user@] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]21:38
-!- cwidmer [8ca3fe9d@gateway/web/freenode/ip.] has quit [Quit: Page closed]23:14
--- Log closed Wed Feb 01 00:00:19 2012