IRC logs of #shogun for Friday, 2013-01-25

--- Log opened Fri Jan 25 00:00:38 2013
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has quit [Quit: Page closed]00:55
-!- FSCV [~FSCV@187.210.54.165] has quit [Quit: Leaving]01:00
-!- cavalier- [cavalier-@70.118.110.73] has quit [Ping timeout: 248 seconds]03:29
shogun-buildbotbuild #261 of nightly_default is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/26103:48
-!- Kawata [~NightKawa@pool-71-126-1-234.bflony.east.verizon.net] has joined #shogun04:53
-!- Kawata [~NightKawa@pool-71-126-1-234.bflony.east.verizon.net] has left #shogun ["Leaving"]04:53
-!- cavalier- [cavalier-@70.118.110.73] has joined #shogun04:55
-!- tomv_ [~tomv@p4FD62165.dip.t-dialin.net] has quit [Ping timeout: 276 seconds]05:00
-!- blackburn [~blackburn@37.61.179.249] has quit [Quit: Leaving.]06:10
-!- blackburn [~lisitsin@mxs.kg.ru] has joined #shogun07:01
-!- blackburn [~lisitsin@mxs.kg.ru] has left #shogun []07:01
-!- blackburn [~lisitsin@mxs.kg.ru] has joined #shogun07:01
blackburnsonney2k: okay I'll take a look on saturday07:17
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun08:25
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has joined #shogun10:18
-!- Hausas [~admin@unaffiliated/hausas] has quit [Ping timeout: 252 seconds]11:10
-!- FSCV [~FSCV@187.210.54.165] has joined #shogun14:09
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has quit [Quit: Page closed]14:25
-!- FSCV [~FSCV@187.210.54.165] has quit [Quit: This computer has gone to sleep]15:18
-!- FSCV [~FSCV@187.210.54.165] has joined #shogun15:20
-!- blackburn [~lisitsin@mxs.kg.ru] has quit [Read error: Connection reset by peer]15:40
-!- jmtrivial [c131a61a@gateway/web/freenode/ip.193.49.166.26] has joined #shogun16:30
-!- jmtrivial is now known as jmfavreau16:30
jmfavreauhi. I'm a new shogun user16:31
-!- Hausas [~admin@185.6.154.4] has joined #shogun16:44
-!- Hausas [~admin@185.6.154.4] has quit [Changing host]16:44
-!- Hausas [~admin@unaffiliated/hausas] has joined #shogun16:44
-!- sonne|work [~sonnenbu@194.78.35.195] has quit [Quit: Leaving.]16:59
jmfavreauis CDenseFeatures has the load function implemented?17:36
jmfavreauor do I have to use the serialized version?17:36
-!- jmfavreau [c131a61a@gateway/web/freenode/ip.193.49.166.26] has quit [Quit: Page closed]18:58
-!- blackburn [~blackburn@85.114.161.134] has joined #shogun20:03
-!- blackburn [~blackburn@85.114.161.134] has quit [Quit: Leaving.]21:07
-!- blackburn [~blackburn@85.114.161.134] has joined #shogun21:13
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has joined #shogun21:20
n4nd0sonney2k: around?21:35
blackburnsonney2k: I am here instead of him :D21:37
blackburnoops21:37
blackburnn4nd0:21:37
blackburn:D21:37
n4nd0blackburn: haha ok21:37
n4nd0so I will fire at you21:37
n4nd0but I already asked you about this the other day, it is same question but rephrased21:38
n4nd0blackburn: do you happen to know how well the structured SVM may work with continuous structured data?21:38
n4nd0think for example of something like an HMM21:38
n4nd0where the states are also real numbers21:39
blackburnahh that time series21:39
n4nd0(i.e. a linear dynamical system)21:39
blackburnI actually think it is kind of wrong21:40
blackburnSVM is here just to estimate dependencies we doesn't happen to know21:40
blackburnif you need parameters it is math21:40
blackburnbut may be I don't get it a little21:41
n4nd0mmm21:50
n4nd0when you say if you need parameters it is math21:50
blackburnyeah I think so21:50
n4nd0blackburn: do you mean it like parameter estimation in the sense of a generative model?21:50
blackburnkind of, yes21:50
n4nd0well but you can use SVM as an alternative to generative models21:51
blackburnI see nothing discriminative in estimation of A,B,C,D of linear system21:51
n4nd0discriminative models more in general21:51
n4nd0but it is discriminative in the sense that if I tell you21:51
n4nd0take a look to these observations21:51
n4nd0now tell me where did they come from, what was the state of the system them21:52
n4nd0there a lot of possible states21:52
n4nd0you find one of them21:52
blackburnyeah but I feel it a little cumbersome21:52
n4nd0hehe I don't understand why :)21:53
blackburncan't prove :D my ass feeling21:53
n4nd0the thing is I am not trying to estimate the parameters A, B, C and D of the system in an SVM fashion21:53
blackburnI don't understand how structured svm works21:53
blackburnyes I understand the math and some cases like multiclass but I don't realize what is it in case of infinite states21:54
n4nd0yeah me neither21:55
n4nd0I don't really know if it can even work21:55
n4nd0let me show you a part of a paper21:55
blackburnI have an feeling that it can be wrong actually :D21:55
blackburnSO comes from crammer and singer formulation with basic generalization21:55
n4nd0http://vision.lbl.gov/Conferences/cvpr/Papers/data/papers/1099.pdf21:56
blackburnI have no reasons left behind to do that, so it is all pretty doubtful for me21:56
n4nd0second page21:56
n4nd0second column21:56
n4nd0they say something like21:56
n4nd0CRFs, SO-SVM and others21:57
n4nd0are for structured prediction case for *discrete* outputs21:58
n4nd0and that there are generatilizations to *continuous* outputs21:58
blackburnright21:58
n4nd0the reference for this one is related to kernel density estimation with joint kernel maps21:58
blackburnn4nd0: just recalled, did I told you I received two mails about tapkee already?21:59
n4nd0and actually the paper ^ with this SOAR thingy looks kind of for this continuous outputs too21:59
n4nd0blackburn: I read something you mentioned in a conversation with Heiko or maybe it was with Viktor21:59
blackburnn4nd0: I should read the paper to understand it more21:59
n4nd0something about a graduate student and later the professor IIRC?22:00
n4nd0blackburn: yeah me too22:00
blackburnn4nd0: I was wrong actually22:00
blackburntotally different people22:00
blackburn:)22:00
blackburnin means of relations22:00
n4nd0blackburn: I have not diffed into these papers yet22:00
blackburndifferent cities even22:00
n4nd0blackburn: haha ok22:00
n4nd0blackburn: that's even better, more users22:00
blackburnn4nd0: have you seen I set up google analytics?22:01
blackburnwe should install it to the shogun page too22:01
n4nd0I saw the fancy github page you sent me22:02
n4nd0but I don't think that is what you meant, is it?22:02
blackburnn4nd0: oops how can you see :D22:03
n4nd0blackburn: what? :)22:03
blackburnn4nd0: I am the only user who sees this I guess22:03
blackburnn4nd0: okay I set up js script on index.html22:03
blackburnthat collects data about visits22:03
n4nd0oh I could see that because you sent me a link yesterday hehe22:05
blackburnn4nd0: link to?22:05
n4nd0the fancy webpage I was talking about22:05
blackburnhttp://lisitsyn.github.com/tapkee/22:05
blackburnthat one?22:05
n4nd0but I think that doesn't have anything to do with google anlytics, not sure though22:05
n4nd0yep22:05
blackburncheck sources22:06
blackburnit has some kind of collecting script22:06
n4nd0mm I don't find what you are saying22:10
blackburnn4nd0: check index.hmtl22:11
blackburnhttps://github.com/lisitsyn/tapkee/blob/gh-pages/index.html22:12
n4nd0I see something with google analytics there22:16
blackburnyeah and I have access to that data22:16
n4nd0blackburn: is it storing in there every time we get a visit?22:16
blackburnyes22:16
n4nd0that's kind of cool :)22:17
n4nd0what do you want to do with that data?22:17
blackburnnothing it is just interesting to see that22:17
n4nd0indeed hehe22:17
n4nd0http://www.google-analytics.com/ga.js22:17
blackburnn4nd0: the page has 14 unique visitors so far22:18
n4nd0that file is quite unreadable for me at least22:20
n4nd0blackburn: where do you see that?22:20
blackburnin the google analytics account22:21
blackburnI am searchng a way to share that somehow22:21
blackburnn4nd0: I've added you22:22
blackburncheck the GA22:23
n4nd0blackburn: I can it too now22:25
n4nd0really cool22:25
n4nd0I can see*22:26
blackburnwe can haz statz22:26
n4nd0so all these statistics are about the fancy webpage22:30
n4nd0not the normal github one, right?22:30
blackburnright22:31
n4nd0blackburn: I am excited about shogun's workshop22:38
blackburnI am too22:38
n4nd0blackburn: are you planning to stay in Berlin for some days?22:39
blackburnn4nd0: yes I hope so22:39
n4nd0cool22:39
blackburnI should find some travel offer22:40
n4nd0sure22:40
n4nd0we will probably know more of it and can start planning better after the meeting tom22:40
blackburncause I don't want to deal with visa :D22:40
blackburnsome travelling company should take care of this I hope22:40
blackburnso the meeting is tomorrow right?22:41
blackburnah I see22:42
blackburnI managed to miss that mail somehow22:42
n4nd0blackburn: there it goes22:45
blackburnn4nd0: hah thanks I meant I just missed it in my box22:45
n4nd0haha whatever22:45
blackburnhmm 20 utc22:45
blackburnso for me it is hmmm22:45
n4nd0here it is 2122:46
n4nd0so there 23 I think22:46
blackburn2422:46
blackburn:D22:46
blackburnrussia is controlled by aliens so +422:47
blackburnthey prefer when sunset happens like 3:20 or so22:47
n4nd0aaah22:48
@sonney2kn4nd0, don't trust the aliens23:03
@sonney2kblackburn, recall that we have analytics for years on shogun-toolbox.org23:03
blackburnsonney2k: really?23:03
@sonney2kyou just didn't want to have access to it...23:03
blackburn:D23:03
blackburnohh I see23:03
blackburnsonney2k: no you always told me you don't have exact stats23:04
blackburnsome misunderstanding happened23:04
@sonney2kblackburn, I have no download stats but all the rest I have23:04
blackburnsonney2k: can you share it?23:04
@sonney2kblackburn, I offered this to you > 1 year ago23:04
blackburnI am constantly drunk come on23:05
blackburnbe merciful23:05
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has quit [Ping timeout: 245 seconds]23:07
blackburnsonney2k: I was young and stupid23:07
blackburn:D23:07
@sonney2kand you didn't need the money I know :D23:07
@sonney2kblackburn, you've got mail23:07
blackburnthanks23:08
blackburnsonney2k: nice load we have23:09
blackburnsonney2k: posting that we use eigen leads to 60 clicks :D23:10
blackburninteresting23:10
blackburnfrom the eigen3 site23:11
blackburnsleep time23:13
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has joined #shogun23:18
n4nd0sonney2k: what aliens? :)23:18
n4nd0tell me about it tomorrow hehe23:27
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has quit [Quit: Page closed]23:27
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 256 seconds]23:58
--- Log closed Sat Jan 26 00:00:38 2013

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!