Open in new window / Try shogun cloud
--- Log opened Sun Aug 19 00:00:17 2012
blackburnsonney2k: I think I will add sparse features with implicit normalization00:02
blackburnand mean -> 000:03
blackburncostly thing however00:03
blackburn[DEBUG] correct=0, total=2150, rejected=000:13
blackburnomg00:13
-!- gsomix [~gsomix@178.45.39.145] has quit [Ping timeout: 245 seconds]01:04
-!- CIA-21 [~CIA@204.152.223.100] has quit [Remote host closed the connection]01:29
shogun-buildbot_build #59 of nightly_none is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/5903:00
shogun-buildbot_build #70 of nightly_default is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/7003:02
shogun-buildbot_build #55 of nightly_all is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_all/builds/5503:03
shogun-buildbot_build #381 of deb1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/38104:26
-!- emrecelikten [~emre@trir-5d800450.pool.mediaWays.net] has quit [Ping timeout: 244 seconds]04:26
-!- emrecelikten [~emre@trir-4d0d87fa.pool.mediaWays.net] has joined #shogun04:40
shogun-buildbot_build #391 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/391  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>04:44
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun07:43
n4nd0good morning07:44
blackburnhey09:31
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has joined #shogun09:44
-!- gsomix [~gsomix@178.45.39.145] has joined #shogun10:53
gsomixhi10:53
wikingho10:59
blackburnn4nd0: so what about risks with hm?11:54
blackburncan I help you?11:54
n4nd0blackburn: I have to take a look to the PR that wiking did11:54
blackburnwell he just put risk to so model11:54
n4nd0for what I undestood11:54
n4nd0it should work with that11:55
n4nd0do you remember the name of the PR?11:57
blackburnnot really11:57
n4nd0has it already been merged?11:57
n4nd0got it11:57
n4nd0https://github.com/shogun-toolbox/shogun/pull/71911:57
blackburnyes sure11:57
n4nd0I am going to issue a new PR soon and later I will check that12:00
n4nd0this is a small one12:00
n4nd0I just added a new Evaluation class, StructuredAccuracy, to compute accuracy for structured problems12:00
blackburngood thing to have12:01
n4nd0definitely12:02
wikingwhat's with the risk fuction then?12:09
wiking*function?12:09
wikingblackburn: merge my last pr please12:10
wikingn4nd0: ping me when u have 10 mins to chat12:10
blackburnwiking: CIA died - it is already merged12:10
n4nd0wiking: now is good, tell me12:10
wikingblackburn: lol ok12:11
wikingn4nd0: ok brb in 1 min12:11
n4nd0ok12:11
wikingn4nd0: i need to redesign or soemting structuredmodel12:11
wikingbut i need your input/consent12:11
wiking:)12:11
n4nd0ok12:11
wikingbtw12:11
wikingcheck my last PR12:11
n4nd0I have not really looked at the PR yet12:11
n4nd0give some more time then12:12
wikingi've change how PSI is calculated in latent12:12
wikingas this is the only way we can support sparse features12:12
wikinghttps://github.com/shogun-toolbox/shogun/pull/74112:12
wikingas the SGVector<float64_t> psi (...) both in Structured and LatentModel12:13
wikingcould not support sparse features12:13
wikingok i'll be back in 1 min12:13
n4nd0wiking: but those changes are just in LatentModel12:14
n4nd0wiking, blackburn: I am checking PR 719, I thought that the idea was to move the risk function to the StructuredModel rather than to the children12:16
n4nd0as I see there it is not implemented in CStructuredModel but in CMulticlassModel12:17
blackburnhow can that be implemented in structured model?12:17
n4nd0do you remember that conversation we had at midnight some days ago?12:18
n4nd0I think that the conclusion there was that the risk function was dependent on the argmax and the psi12:18
wikingn4nd0: for now yes12:18
wikingn4nd0: that risk function only valid for Multiclass case12:18
wikingthat's why it's in multiclass12:18
n4nd0but apart from that, the definition is genereal12:18
blackburnn4nd0: no, subgrad is not general12:18
wikingn4nd0: the risk fucntion for instance for SOlatent svm would be different12:19
n4nd0blackburn: wasn't it too in the definition we saw in that thesis?12:19
n4nd0wiking: that's why I think it should be virtual, but in StructuredModel12:19
blackburnyes, for some models it is12:19
shogun-buildbot_build #392 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/392  blamelist: Viktor Gal <viktor.gal@maeth.com>12:19
wikingn4nd0: and about the psi function: i've just done it in latent model since i don't know how it would effect SO at the moment12:19
n4nd0wiking: ok12:19
wikingn4nd0: it is virtual in structuredModel12:20
n4nd0wiking: let me show you the reference we were talking about12:20
n4nd0well I think there is a mail with that12:20
wikingn4nd0: https://github.com/shogun-toolbox/shogun/pull/719/files#L18R22812:20
blackburnn4nd0: I don't mind it to be general12:20
blackburnbut if there is loss of efficiency I don't mind to have some duplicated code12:21
wikingsomebody broke java modular :)12:21
wikingi guess it's heiko12:21
n4nd0http://users.cecs.anu.edu.au/~chteo/pub/Teo10_phdthesis.pdf12:21
blackburnI mean if {argmax+subgrad} in multiclass is more efficient than virtual one - we should go this way12:22
blackburnsee what I mean?12:22
n4nd0blackburn: why should it be more efficient that way?12:22
wikingn4nd0: which page?12:22
blackburnwiking: Appendix 112:22
n4nd0wiking: equations A.1 and A.8, pages 103 and 10512:22
n4nd0they are the contribution in the risk of one training example and the subgradient, respectively12:23
wikingok12:23
wikingbut isn't this application specific?12:24
wikingi mean what i understood from uricamic's solver12:24
wikingit pretty much lambda*||w||_2^2 + Risk12:24
wikingi mean it solves this12:24
blackburnhahaha I just noticed the author's name12:25
wikingand as you can see from the risk function of MulticlassModel... it pretty much works only if your SO problem is a multiclass problem12:26
blackburnbad bad word in the middle12:26
n4nd0wiking: aren't the application speficic parts only in the argmax and the psi?12:26
wikingn4nd0: in that risk funciton no12:26
wiking*function12:27
blackburnn4nd0: so you want to implement subgrad in independent fashion as well12:27
blackburnright?12:27
wikingit loops through the various classes12:27
n4nd0wiking: in uricamic's one?12:27
wikingn4nd0: yes12:27
n4nd0wiking: can you point out for me the loops you mean? lines?12:27
wikingline 18112:28
wikingin MulticlassModel.cpp12:28
n4nd0wiking: it looks to me that this part is the argmax12:28
wikingwell if yes then chuck it out :)12:29
wikingand replace it with argmax :)12:29
wikingand see if we can get further with the story12:29
n4nd0hehe I guess I should read it completely first12:29
wikingand move the risk function in structredmodel ;)12:29
wikingR({\bf w}) = \sum_{i=1}^{m} \max_{y \in \mathcal{Y}} \left[ \ell(y_i, y) * + \langle {\bf w}, \Psi(x_i, y) - \Psi(x_i, y_i)  \rangle  \right]12:30
wikingthis is what the code does apparently12:30
wiking(see the doxygen for virtual float64_t risk(float64_t* subgrad, float64_t* W, TMultipleCPinfo* info=0); in MulticlassModel.h12:30
wikingbut let me know when u r through this one as i have other concerns :)12:31
n4nd0tell me about them :)12:32
wikingheheh so ok the following is the story12:32
wikingi need to be able to chuck in extra variables for StructuredModel12:32
wikingfor delta_loss12:32
wikingand get_joint_vector...12:32
wikingwhich would mean that i need to change the function arguments in StructuredModel12:33
wikingbut that won't work  :)12:33
wikingi mean i need something like this for example for delta loss12:33
wikingloss(y, yhat, hhat) instead of loss(y1, y2)12:34
n4nd0ok12:34
n4nd0I thought of that12:34
wikingget_joint_feature_vector should be12:35
n4nd0my idea was to put that method in StructuredModel too12:35
wikingget_joint_feature_vector(feat_index, structureddata y, latentdata h)12:35
blackburnn4nd0: btw about coffin integration12:35
blackburnsubgrad of models (by Teo)12:35
blackburnis 0 + \phi(x,\overline{y})12:36
n4nd0wiking: just put them in the StructuredModel12:36
blackburn- \phi(x,y)12:36
wikingn4nd0: how? :)12:36
blackburnI mean it is two add_to_dense_vec operations12:36
wikingi can do something like: get_joint_feature_vector(feat_index, structureddata y, latentdata h = NULL)12:36
n4nd0wiking: that's why I didn't do delta_loss and get_joint_feature_vector pure virtual, but unimplemented12:36
wikingand same for delta_loss12:36
n4nd0exactly, that's the spirit :)12:36
wiking?12:36
wikingno comprende12:36
n4nd0mmm12:37
n4nd0what is the problem of making new methods?12:37
n4nd0we have a delta_loss(y1,y2)12:37
n4nd0put another one there delta_loss(y,yhat,hhat)12:37
blackburnwiking: can you convert it to sth like 'add_joint_feature_vector_to_dense_vec'? :)12:37
shogun-buildbot_build #393 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/393  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>12:38
wikingblackburn: which where?12:38
blackburnwiking: I mean now you are using get_joint_feature_vector12:38
wikingn4nd0: and who will call it in your Primal solver?12:38
blackburnoh I will wait you to finish :)12:38
wikingblackburn: which class r u talking about12:39
blackburnwiking: let me check last pr again12:39
n4nd0wiking: aham, so you would like to use the primal solver for your latent problems directly12:40
wikingyes12:40
wikingas well as the dual solver by uricamic12:40
n4nd0wiking: aham, are you sure it can be done like that?12:40
wikingwell the only difference is12:40
wikinghow you calculate the delta loss and the psi12:40
wikingi.e. psi is based on x,y,h12:41
wikingand not only on x,y12:41
wikingsame with the delta loss12:41
n4nd0ok12:41
wikingsee here:12:41
n4nd0but even if you include a new parameter latentdata h = NULL12:41
n4nd0how do you plan to make the difference between executing for latent or not?12:42
n4nd0some kind of boolean member?12:42
wikingn4nd0: no12:42
wikingi mean from the fucntion you can tell whether the func arg is set or not12:42
wikingright?12:42
wikingbtw i can see that you use only argmax directly from PrimalMosek12:43
wikingright?12:43
n4nd0yes, I think so12:43
n4nd0wiking: and yes, you are right about the thing of the func arg; I am king of retarded...12:44
n4nd0kind* of12:44
wikingmmm12:44
wikingthen maybe this would work12:44
wikingbut the stupid thing is12:44
wikingmmm12:44
wikingsomehow i wanted to hide this behind the function if you know what i mean12:45
wikingso that things are not passed via the function arg12:45
wikingrather you just give an index or something12:45
wikingand everything else is behind the scenes12:45
wikingblackburn: ok check it again12:46
wikingn4nd0: but maybe it'll work what i'm thinking of now12:46
blackburncheck what?12:46
wikingas really delta_loss and get_joint_feature_vector is being called in your argmax12:46
wikingblackburn: last pr12:46
wikingblackburn: you were about to comment there something12:47
blackburnI'm kind of lost there12:47
wikingblackburn: this one: https://github.com/shogun-toolbox/shogun/pull/74112:48
wiking?12:48
wikingin this one?12:48
blackburnno in general12:48
blackburnso you compute dense features12:48
blackburnaccording to position of window12:48
blackburnright?12:48
wikingblackburn: mmm don't mix it now with the application :)))12:48
blackburnbut is it so?12:49
wikingblackburn: yeah there is so like that12:49
wiking*it is like that12:49
blackburnit sounds like memory eating thing12:49
n4nd0wiking: ok, so tell me if so when you have ready with what that you were thinking of12:50
wikingn4nd0: well now i'm thinking of creating a class that is actually a direct inheritance of StructuredModel12:50
wikingn4nd0: and add there non-virtual functions for delta_loss and get_psi12:51
wikingn4nd0: and the argmax and basically then the argmax will call the right functions with the right amount of arguments12:51
wikingand then for now i don't touch the StructuredModel12:51
wikingand if that one works out then we'll see later about how we could simplify the thing12:52
wikingi have no better idea now :S12:52
n4nd0it looks like a good idea to me12:52
n4nd0probably those two classes could be merged once we check it works12:52
n4nd0there the PR goes :)13:31
blackburn??????13:43
blackburnmerged13:43
n4nd0thank you :)13:44
wikingcia bot die!13:58
wikingand it did apparently13:59
shogun-buildbot_build #394 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/394  blamelist: iglesias <fernando.iglesiasg@gmail.com>14:08
blackburnn4nd0: did you read about CRFs before? is it's argmax hard to implement?14:27
n4nd0blackburn: I have read something about CRFs ues14:28
n4nd0yes*14:28
n4nd0I don't think the argmax is something fancier than Viterbi14:28
blackburnI was pretty excited when I saw CRF based segmentation14:29
n4nd0BUT the thing is that I don't know how CRFs would fit into our SO framework14:29
n4nd0why?14:29
blackburnwhy not14:29
blackburn?14:29
blackburnwhat is so different between HM and CRFs?14:29
n4nd0in the paper I read the way of training had nothing to do with the one for HM-SVM I know14:30
blackburnhmm I see14:32
shogun-buildbot_build #395 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/395  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>14:33
n4nd0blackburn: the build was already broken before right?14:34
blackburnyes14:34
blackburnby heiko14:34
wikingneeeeeeedcoffee14:49
-!- heiko [~heiko@host86-185-9-87.range86-185.btcentralplus.com] has joined #shogun14:50
wikingheiko: broker! :) wazza?15:04
heikowiking hello15:04
heikoehm15:05
heikoI didn't do it :D15:05
wikingCRF CRF blackburn will u implement? :)15:05
-!- pluskid [~pluskid@76.79.79.246] has joined #shogun15:15
shogun-buildbot_build #396 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/396  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>15:34
-!- pluskid [~pluskid@76.79.79.246] has quit [Ping timeout: 276 seconds]15:34
heikoshogun-buildot_ status15:49
heikowiking!15:51
heikounit-tests!15:51
heikostill dont work15:51
heikogtest refuses to install15:51
heikohow to I do this?15:51
shogun-buildbot_build #397 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/397  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>15:52
wikingheiko: ahahhaha yeah i told u to ask blackburn16:00
wikingheiko: i dont know wtf is with that shit but the best is that you do the following16:01
wikingsudo cp lib/.libs/libgtest.so* /usr/local/lib16:01
wikingsudo cp lib/.libs/libgtest_main.so* /usr/local/lib16:01
wikingheiko: and then the configure should work finally16:01
wikingi dont know why do they do this at google... i mean to refuse to do a make install :)16:03
shogun-buildbot_build #398 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/398  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>16:05
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has quit [Ping timeout: 265 seconds]16:16
shogun-buildbot_build #399 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/399  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>16:23
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has joined #shogun16:26
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has quit [Ping timeout: 260 seconds]16:33
n4nd0 /exit17:14
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]17:14
-!- wubino [~wubino@74.61.91.195] has joined #shogun17:47
-!- puffin444 [180317bb@gateway/web/freenode/ip.24.3.23.187] has joined #shogun17:58
puffin444hey blackburn17:59
-!- heiko [~heiko@host86-185-9-87.range86-185.btcentralplus.com] has left #shogun []18:18
puffin444hey blackburn, just to let you know, i rebased my repo so its ready to be merged18:20
blackburnpuffin444: hey18:23
blackburnokay nice18:23
-!- heiko [~heiko@host86-185-9-87.range86-185.btcentralplus.com] has joined #shogun18:32
-!- heiko [~heiko@host86-185-9-87.range86-185.btcentralplus.com] has quit [Client Quit]18:35
blackburnwiking: I will if n4ndo not18:40
shogun-buildbot_build #289 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/289  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>18:42
-!- wubino [~wubino@74.61.91.195] has left #shogun ["Leaving"]19:23
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has joined #shogun19:50
wikingblackburn: :D20:58
-!- os253 [~os@e180116168.adsl.alicedsl.de] has joined #shogun20:59
os253Hi puffin44421:00
puffin444hello os25321:00
os253is now good for you?I am around all evening.21:01
puffin444yes now is good if its good for you21:01
blackburnoh gaussian team in da house21:01
puffin444yes indeed :)21:02
os253yes, fine here.21:02
os253:-)21:02
blackburntell me - how much effort is to add GP classifiers?21:02
puffin444honestly not sure. I would have to check GPML.21:02
puffin444Shouldnt require too much though21:03
os253How far is the Laplace code? If that works it's just another likelihood function...21:03
puffin444The laplace code is pretty much done and ready to be merged wt any time.21:03
blackburnabout your PR21:04
puffin444yes?21:04
blackburnpuffin444: you merged it in some kind of broken way - can you clean it up?21:04
puffin444uh oh what's the problem?21:05
blackburnos253: have you seen puffin444 PR? https://github.com/shogun-toolbox/shogun/pull/70721:05
blackburnpuffin444: latest commits include some heiko's ones..21:05
blackburnhowever may be it is not worth it21:05
os253Then it should be easy. blackburn: do you think classification is useful? Generaly speaking: gps are not that great for classifcation or at least not better than SVMs.21:05
blackburnos253: I am pretty lame in everything about GPs :)21:06
blackburnos253: I am kind of chris widmer representer :D21:07
os253i see21:07
puffin444os253 all the main features mentioned in the project plan have been merged or are ready to be merged21:09
blackburnpuffin444: can you change namespace in brent method source?21:09
puffin444At this point any additional feature should be a fairly easy task21:09
blackburnpuffin444: I have two general issues I think would be cool to solve21:11
blackburnfirst is rather mandatory - FITC eigen members21:11
puffin444yes?21:11
blackburnand second is some alternative to nlopt in gradient model selection21:11
blackburnno idea if you have time for that :)21:11
puffin444Well gradient model selection is pretty general21:12
blackburneigen member is serious issue we need to resolve before release21:12
blackburnmembers*21:13
puffin444What about FITC eigen membes? Is that the map<vecrorxd> we discussed earlier?21:13
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun21:13
blackburnyes21:13
blackburnso please focus on that before 1st of sept21:13
puffin444okay. So replace those Vectorxd members with map?21:13
blackburnno, replace it with SGMatrix and SGVector21:14
blackburnand map in your code21:14
os253what is the issue with eigen members?21:15
blackburnos253: eigen should not be included and/or exposed in headers to avoid additional dependency21:15
puffin444in .h or also .cpp?21:15
blackburnpuffin444: I think I have an idea21:15
blackburnI am going to add as_eigen method21:15
blackburnto both sgvector and sgmatrix21:16
os253I see21:16
blackburnwhich returns eigen analogue of the matrix21:16
blackburnor the vector21:16
blackburnsee what I mean?21:16
puffin444yes.21:16
blackburnoops not possible21:16
blackburnsame reason21:17
blackburn:D21:17
blackburnI missed that21:17
blackburnyou have to map it manually21:17
blackburnokay just change it to sg* and map in your code before use21:17
puffin444Okay. Can I still use eigen vectors directly in cpp files?21:17
blackburnyes sure21:18
puffin444os253 I am working ona tutorial pagefor gps21:19
os253yes, that would be great. I have looked through the PR yesterday. It seems like a good mix of sticking to the GPML code and adapting it to your framework.21:20
blackburnpuffin444: Map<MatrixXd> (m_some_matrix.matrix,m_some_matrix.num_rows,m_some_matrix.num_cols);21:20
blackburnthat's what you would have to do21:20
os253I am wondering how to best make sure it will be picked up. What examples do you currently have?21:20
blackburnerr missed name21:20
os253It looks like tutorials and examples are the main thing to add.21:20
puffin444I have examples for every msjor feature under libshogun21:20
os253Perhaps the21:20
puffin444I also have a basic graphical example in python21:21
puffin444All the main features have been integrated into the modular interface21:21
os253I'd vote for a python example, just a single one where you use standard GP, GP with ficti (also visusalizing the location of the learnt inducing points).21:22
blackburnpuffin444: btw are you going to research GPs further?21:22
os253Lastly the same with student-T. Something like that.21:23
puffin444for graduate school?21:23
blackburnno idea, in general21:23
puffin444I hope to stay and contribute to shogun and flesh the gp code out a bit21:24
blackburncool21:24
puffin444os253 that should beeasyto do.21:24
os253 puffin444: you almost should stay on, having coded up this chunk is substantial. From that basis it would be easy to carry out novel directions that are more research like.21:25
os253Great, I think a visual python example would be very useful to have.21:25
puffin444Yes I could just build onto the current one I have.21:26
puffin444blackburn, is it possible that the current pr be merged? The only reason I ask is that I have to fix a ton of conflicts each time I have to rebase.21:29
os253that would be great.21:29
blackburnpuffin444: ah sure I forgot21:29
blackburnmerged21:30
blackburnCIA died so no message21:30
puffin444os253, Do you have any orther questions/concerns?21:31
os253I have a list of small things, but perhaps I can just email them. For example: choice of the number of inducing points.21:32
os253You currently take the same approach than GPML, right? I.e. they are fixed.21:34
puffin444fixed at what point? They are chosen before hyperparameter learning,21:35
os253Or do you have code for choosing an apprpriate number? A simple heuristic may be good enough.21:35
puffin444Oh yes in that case they are fixed.21:35
shogun-buildbot_build #389 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/389  blamelist: puffin444 <walke434@msu.edu>21:36
os253Perhaps one could add this to a python demo for large data. Do you have a suitable datasets with thousands of data points to illustrate the use of the sparse approximation?21:36
shogun-buildbot_build #390 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/390  blamelist: puffin444 <walke434@msu.edu>, root <root@gs12672-vm.sp.cs.cmu.edu>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>, puffin444 <walke434@gmail.com>21:36
puffin444No I do not.21:37
os253What datasets are you using at the moment?21:38
puffin444A simple 4 point dataset I use to compare with gpml.21:38
os253Ok. I can send you some suggestion.21:39
puffin444I did this to ensure the hyperparameter learning came up with consistent parameters across gpml and shogun.21:39
puffin444yes.21:39
os253Perhaps best to pick some one can download from a script to avoid having to add it to the repository.21:40
n4nd0blackburn: hey there21:40
os253It would be useful to have a script that can be run, downloads the datasets and processes it alongside some plots.21:40
n4nd0blackburn: when someone publishes a paper, don't you think that the code and data they have used for their results should be provided?21:41
os253I'll send you a list of datasets that are public and can be downloaded.21:41
puffin444I could try that.21:41
n4nd0blackburn: I am quite surprised that I don't manage to find anything about it for the papers I am reading, maybe I am searching wrong :S21:41
puffin444thanks21:41
-!- gsomix_ [~gsomix@178.45.41.90] has joined #shogun21:42
os253Yes, that would be great. Thanks. I'll send this by tomorrow.21:43
-!- gsomix [~gsomix@178.45.39.145] has quit [Ping timeout: 244 seconds]21:45
os253If you are around tomorrow we can briefly chat then.21:46
os253I am online all day but not necessarily on IRC.21:46
puffin444sorry blackburn. It looks like I forgot a HAVE_EIGEN somewhere21:46
blackburnn4nd0: yeah kind of open source science21:46
blackburnI believe most cases of not available code means NO CODE21:47
blackburnand no research21:47
n4nd0blackburn: but how can it be possible that there is no code?21:47
blackburnI mean it could turn to be a false research21:47
n4nd0I mean, the results that are published must be checked somehow!21:47
blackburn:D21:47
puffin444os253, I unfortunately will not be available tomorrow.21:47
blackburnno they are not checked by editors21:47
n4nd0blackburn: really?21:47
blackburnn4nd0: well they are checked for consistency or so21:48
blackburnbut nobody will try to run your code21:48
os253puffin444: ok -then let's do via email and or at a later time.21:48
puffin444ok.21:49
n4nd0blackburn: I am really surprised21:49
n4nd0when I start writing papers21:49
n4nd0I will offer a link to the code!21:49
n4nd0hehe21:49
blackburnyeah that's how it should be21:50
blackburnpuffin444: can you fix it?21:50
puffin444I will try.21:50
blackburnokay thanks21:50
puffin444Can't do it immediately, but i can to it in a few hours21:51
blackburnI will do then21:51
blackburnpuffin444: oh.. you added new methods with eigen return type21:52
puffin444uh oh. How can i resolve it?21:53
blackburnthe same - return sgvectors/sgmatrices21:54
blackburnand then map21:54
puffin444ok. will do.21:55
blackburnI wish we had some solution for that but I think depending on eigen is wrong21:55
puffin444ok.21:56
os253blackburn: the more code requirew21:57
os253*requires Eigen the more dominant implicit dependencies become anyway. At some point it will make little difference :-)21:58
blackburnos253: yeah that's clear21:58
blackburnI have mixed point on that..21:58
blackburnat some point we won't be able to avoid that21:59
os253true...22:01
shogun-buildbot_build #290 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/290  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>22:03
blackburnshogun-buildbot_: wait a minute wait a minute22:09
puffin444blackburn, should I try to fix the buid in a few hours or are you doing it?22:09
blackburnpuffin444: I am doing22:10
puffin444ok.22:10
puffin444Then i will get to work on the mao stuff22:10
puffin444map*22:10
shogun-buildbot_build #391 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/391  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>22:20
n4nd0good night people22:43
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]22:44
-!- os253 [~os@e180116168.adsl.alicedsl.de] has quit [Remote host closed the connection]22:52
-!- blackburn [~blackburn@62.106.106.114] has quit [Quit: Leaving.]23:01
-!- blackburn [~blackburn@62.106.106.114] has joined #shogun23:02
-!- blackburn1 [~blackburn@31.28.44.3] has joined #shogun23:05
-!- blackburn [~blackburn@62.106.106.114] has quit [Ping timeout: 252 seconds]23:06
-!- blackburn2 [~blackburn@109.226.112.41] has joined #shogun23:06
-!- CIA-36 [cia@198.71.88.9] has joined #shogun23:09
-!- blackburn1 [~blackburn@31.28.44.3] has quit [Ping timeout: 240 seconds]23:09
shogun-buildbot_build #392 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/392  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>23:15
shogun-buildbot_build #393 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/393  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>23:20
blackburn2mother fucker23:26
blackburn2will that end?23:26
blackburn2shogun-buildbot_: you are killing me bastard23:35
blackburn2puffin444: I have to ifdef just everything23:37
shogun-buildbot_build #394 of deb1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/39423:45
shogun-buildbot_build #315 of deb2 - static_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/315  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>23:48
shogun-buildbot_build #291 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/291  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>23:51
--- Log closed Mon Aug 20 00:00:17 2012