--- Log opened Mon Jul 15 00:00:32 2013 | ||
thoralf | Bye. | 00:09 |
---|---|---|
-!- thoralf [~thoralf@24-134-81-194-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 00:09 | |
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has joined #shogun | 00:55 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 00:55 | |
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving] | 01:50 | |
shogun-buildbot | build #458 of nightly_default is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/458 | 04:27 |
-!- gsomix [~gsomix@185.6.239.196] has quit [Quit: Leaving] | 05:22 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 05:49 | |
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 256 seconds] | 06:01 | |
-!- nube [~rho@116.90.239.13] has joined #shogun | 06:15 | |
-!- foulwall [~user@2001:da8:215:6100:c01b:d373:53b1:bd42] has joined #shogun | 06:17 | |
-!- foulwall [~user@2001:da8:215:6100:c01b:d373:53b1:bd42] has quit [Ping timeout: 245 seconds] | 06:30 | |
-!- foulwall [~user@2001:da8:215:6901:452b:8de3:b659:6495] has joined #shogun | 06:38 | |
-!- foulwall` [~user@2001:da8:215:503:8587:b3a2:46ae:c7d6] has joined #shogun | 06:40 | |
-!- foulwall [~user@2001:da8:215:6901:452b:8de3:b659:6495] has quit [Ping timeout: 245 seconds] | 06:42 | |
-!- nube [~rho@116.90.239.13] has quit [Ping timeout: 260 seconds] | 08:04 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 08:18 | |
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 256 seconds] | 08:58 | |
-!- nube [~rho@116.90.239.13] has joined #shogun | 09:17 | |
-!- lambday [67157d36@gateway/web/freenode/ip.103.21.125.54] has joined #shogun | 09:18 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 264 seconds] | 09:21 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 09:35 | |
-!- lambday [67157d36@gateway/web/freenode/ip.103.21.125.54] has quit [] | 09:51 | |
-!- lambday [67157e4f@gateway/web/freenode/ip.103.21.126.79] has joined #shogun | 10:24 | |
-!- foulwall` [~user@2001:da8:215:503:8587:b3a2:46ae:c7d6] has quit [Ping timeout: 245 seconds] | 10:45 | |
-!- foulwall` [~user@2001:da8:215:503:801e:26a9:1a95:4b99] has joined #shogun | 10:58 | |
@wiking | sonney2k: ping | 11:04 |
-!- vgorbati [c3ee5cb1@gateway/web/freenode/ip.195.238.92.177] has joined #shogun | 11:10 | |
@sonney2k | wiking, hey there! | 11:16 |
@sonney2k | foulwall`, neat! so the whole machinery beneath is already working? | 11:18 |
@wiking | sonney2k: can we fix buildbot nightly? :) | 11:18 |
@sonney2k | foulwall`, just the centering / extracting the region is missing and I guess you should also load the model just once for this demo if it takes too long | 11:19 |
@sonney2k | wiking, what's wrong with it? | 11:19 |
@wiking | sonney2k: i mean to generate the .tar.gz of the binaries? | 11:19 |
-!- van51 [~van51@athedsl-399972.home.otenet.gr] has joined #shogun | 11:19 | |
@sonney2k | wiking, we just need a make target for that and then we can do it | 11:20 |
@wiking | ok | 11:20 |
@sonney2k | wiking, did sergey come to the 3nd workshop day btw? | 11:20 |
@wiking | btw are we doing this for the r_modular: PKG_LIBS="example.cxx" R CMD SHLIB -o example example_wrap.cpp | 11:20 |
@wiking | " R is also sensitive to the name of the file extension in C and C++ mode. In C++ mode, the file extension must be .cpp rather than .cxx for the R compile command to recognize it. If your C++ code is in a file using something other than a .cpp extension, then it may still work using PKG_LIBS: | 11:20 |
@wiking | swig" | 11:20 |
@sonney2k | wiking, no we are doing something totally custom | 11:21 |
@wiking | sonney2k: oh i see | 11:21 |
@sonney2k | wiking, I did a bit of R debugging in the last days | 11:22 |
@sonney2k | wiking, at least my minimal example worked with ref/unref jsut fine | 11:22 |
@sonney2k | wiking, now I am trying to get a minimal shogun class to work | 11:22 |
@sonney2k | say BinaryLables | 11:22 |
@sonney2k | *Labels | 11:23 |
@sonney2k | if this works then it is just some shogun thing | 11:23 |
@sonney2k | but it is complex... | 11:23 |
foulwall` | ok, sonney2k, gotcha, will send a pr soon. | 11:35 |
-!- thoralf [~thoralf@enki.zib.de] has joined #shogun | 11:38 | |
thoralf | Hey. | 11:38 |
thoralf | Someone could change the topic. ;) | 11:38 |
-!- van51 [~van51@athedsl-399972.home.otenet.gr] has left #shogun ["PING 1373881241"] | 11:40 | |
-!- van51 [~van51@athedsl-399972.home.otenet.gr] has joined #shogun | 11:41 | |
-!- ChanServ changed the topic of #shogun to: "Shogun Machine Learning Toolbox | Devs are usually around between 10am-10pm UTC | GSoC 2013 is running http://bit.ly/16LaK5Z | When asking questions please be patient or use the mailing list. We always answer but it might take a while. | Channel logs http://bit.ly/eQDk8Q" | 11:46 | |
@sonney2k | wiking, did you have the chance to check the videos of the workshop? | 11:48 |
lambday | sonney2k: hi | 11:51 |
@sonney2k | lambday, hey there! | 11:53 |
lambday | sonney2k: I need to add a different dense_dot in SGSparseVector, something that takes another template argument... say, when I need to compute sparse-matrix-vector product when matrix is complex64 and vector is float64 | 11:54 |
lambday | sonney2k: would that be okay? | 11:55 |
@sonney2k | lambday, it makes sense to have these | 12:02 |
lambday | sonney2k: okay... so adding a few of them, complex64, float64, int etc.... | 12:02 |
@sonney2k | lambday, I wanted to have this in CDotFeatures but IIRC deriving from a templated class can be problem | 12:03 |
@sonney2k | a | 12:03 |
thoralf | sonney2k: I thought you'll in holiday now? | 12:05 |
thoralf | you're | 12:05 |
lambday | sonney2k: umm.. problem as in? | 12:05 |
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has joined #shogun | 12:09 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 12:09 | |
thoralf | Hey iglesiasg | 12:09 |
@iglesiasg | good morning! | 12:09 |
van51 | hey | 12:10 |
van51 | sonney2k: I've sent a PR, when you have time please review it :) | 12:10 |
van51 | sonney2k: I've started working on the streaming version of those | 12:10 |
van51 | sonney2k: if you want to discuss anything I'll be around | 12:11 |
van51 | lambday: I think it will be a bit messier to extend a templated class | 12:11 |
@iglesiasg | thoralf: I will fix the SO multiclass example ASAP (I want to finish an LMNN thing first). But if you want to do it, feel free ;) | 12:11 |
van51 | lambday: I tried it two-three days ago with DenseFeatures (which are templated) | 12:11 |
thoralf | iglesiasg: No worries. | 12:11 |
van51 | lambday: ofc I'm not a c++ expert (yet :P) so that may be we it troubled me | 12:11 |
van51 | why* | 12:12 |
lambday | van51: yes templates are real pain :( | 12:12 |
lambday | van51: well I am thinking of having something like this | 12:12 |
thoralf | iglesiasg: I need multilabel instead of multiclass. So I anyway will write my own example/classes. | 12:12 |
lambday | template<class T1> class { template<class T2> void foo(T2 a); }; | 12:13 |
van51 | lambday: yea I saw what you wanted to do, with the sparse product | 12:14 |
lambday | oops I forgot the class name itself :D | 12:14 |
@iglesiasg | thoralf: Aham, I understand. | 12:14 |
lambday | van51: yes | 12:14 |
van51 | lambday: i think it will be really useful :) | 12:14 |
lambday | van51: I hope so.. I will be using those anyway | 12:14 |
thoralf | iglesiasg: I spend a night reading your SO stuff and I think I need some modifications for multilabel. | 12:15 |
lambday | by the way, how did the workshop go, guys? | 12:15 |
lambday | saw the pics :) | 12:15 |
-!- nube [~rho@116.90.239.13] has quit [Quit: Leaving.] | 12:16 | |
van51 | and what about the videos? :D | 12:16 |
@iglesiasg | thoralf: So, first of all you need a new model, right? In multilabel you want to model dependencies between all pairs of classes, or not? | 12:16 |
lambday | yes yes what about the video :D | 12:16 |
@iglesiasg | all pairs of labels, better said probably | 12:16 |
lambday | umm... make unit-tests now treats all warnings as errors :-/ | 12:17 |
@iglesiasg | lambday: the workshop was pretty cool | 12:18 |
@iglesiasg | lambday: I enjoyed a lot the days in there. We had a couple of coding days during the weekend | 12:18 |
lambday | iglesiasg: awesome, man!! | 12:18 |
@iglesiasg | van51: I think wiking or sonney2k are editing them and later they will be uploaded to youtube | 12:19 |
van51 | iglesiasg: ok cool! thanks for letting me know | 12:19 |
@iglesiasg | lambday, van51: did you guys follow the live stream? | 12:20 |
lambday | iglesiasg: no I couldn't :( | 12:20 |
lambday | iglesiasg: will the slides be available? | 12:20 |
@iglesiasg | lambday: I thought they already were | 12:26 |
@iglesiasg | lambday: have you checked the logs? I think there were some links to them pasted here | 12:26 |
@iglesiasg | lambday: anyway, yes. I think they will be uploaded somewhere in the website once the videos are up and so on. | 12:27 |
lambday | iglesiasg: alright.. thanks.. I'll check it out :) | 12:32 |
thoralf | iglesiasg: No, only between some classes. | 12:54 |
thoralf | iglesiasg: It's a taxonomy. | 12:54 |
thoralf | iglesiasg: I don't even need to learn label-label interactions. | 12:55 |
@iglesiasg | thoralf: Aham! Got it. | 12:58 |
@iglesiasg | thoralf: So you could as well use directly multiclass classification too. Each class corresponding to a particular choice of the labels. | 12:59 |
thoralf | iglesiasg: How do I model that a class belongs to class i and class j? | 13:00 |
thoralf | iglesiasg: btw.: I have 7000 labels - so "label combinations" won't work. ;) | 13:01 |
@iglesiasg | thoralf: yeah, indeed that's way too many classes for doing that | 13:02 |
@iglesiasg | thoralf: so if I understand correctly, for your labels you can probably use the Sequence labels, and for the features probably MatrixFeatures work well. What needs to be new is the StructuredModel | 13:04 |
-!- nube [~rho@49.244.49.62] has joined #shogun | 13:18 | |
-!- gsomix [~gsomix@83.234.54.35] has joined #shogun | 13:22 | |
thoralf | iglesiasg: Something like this, yes. | 13:41 |
-!- foulwall` [~user@2001:da8:215:503:801e:26a9:1a95:4b99] has quit [Ping timeout: 245 seconds] | 13:53 | |
thoralf | Somebody who could merge my PR? https://github.com/shogun-toolbox/shogun/pull/1243 | 14:25 |
-!- flxb_ [~flxb@master.ml.tu-berlin.de] has quit [Quit: Lost terminal] | 14:39 | |
@iglesiasg | thoralf: it seems there is some conflict with it, right? | 14:56 |
thoralf | iglesiasg: No, travis wasn't triggered. | 14:57 |
thoralf | I don't know why. | 14:57 |
@iglesiasg | thoralf: but the PR cannot be merged from the github's interface | 14:58 |
@iglesiasg | thoralf: "We can't automatically merge this pull request. Use the command line to resolve conflicts before continuing." | 14:58 |
thoralf | iglesiasg: That's weird. | 14:59 |
thoralf | I'll have a look. | 14:59 |
@iglesiasg | thoralf: I have not seen it before either. I would try to rebase to latest develop and see if I get any conflict | 14:59 |
thoralf | iglesiasg: Okay, I did a rebase and pushed it again (there was indeed a conflicting change) - let's hope github is able to merge it. | 15:07 |
thoralf | iglesiasg: Travis runs, a good sign. | 15:07 |
@iglesiasg | thoralf: good | 15:11 |
@iglesiasg | thoralf: I am no real Makefile expert though, so I am probably not the right one to merge this. Do you need to get this merged soon? | 15:13 |
thoralf | iglesiasg: I guess this happened because HeikoS PR got merged before mine - and introduced a conflict, which wasn't there before. ;) | 15:13 |
@iglesiasg | probably | 15:13 |
thoralf | iglesiasg: Not really, but later merges mean more potential conflicts. ;) | 15:13 |
thoralf | iglesiasg: I can explain my changes - would that help? | 15:14 |
@iglesiasg | thoralf: I understand what's going on. But I don't know if it can potentially arise conflict with sometihng | 15:18 |
@iglesiasg | lunch time now! | 15:18 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 16:02 | |
@iglesiasg | re | 16:06 |
gsomix | today is monday... | 16:17 |
gsomix | hello all | 16:18 |
@iglesiasg | hello hello | 16:18 |
@iglesiasg | gsomix: what's wrong with Monday man? :) | 16:18 |
gsomix | iglesiasg, report time! :) | 16:18 |
@iglesiasg | oh yeah! | 16:19 |
-!- van51 [~van51@athedsl-399972.home.otenet.gr] has quit [Quit: Leaving.] | 16:49 | |
-!- gsomix [~gsomix@83.234.54.35] has quit [Ping timeout: 264 seconds] | 16:51 | |
-!- foulwall [~user@2001:da8:215:c252:19b6:f7da:70f4:e333] has joined #shogun | 16:59 | |
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving] | 17:05 | |
-!- vgorbati [c3ee5cb1@gateway/web/freenode/ip.195.238.92.177] has quit [Quit: Page closed] | 17:24 | |
-!- gsomix [~gsomix@188.168.3.86] has joined #shogun | 17:42 | |
-!- gsomix [~gsomix@188.168.3.86] has quit [Read error: Connection reset by peer] | 17:44 | |
-!- gsomix [~gsomix@188.168.3.86] has joined #shogun | 17:44 | |
-!- foulwall [~user@2001:da8:215:c252:19b6:f7da:70f4:e333] has quit [Ping timeout: 245 seconds] | 18:13 | |
-!- foulwall [~user@2001:da8:215:c252:19b6:f7da:70f4:e333] has joined #shogun | 18:15 | |
-!- foulwall [~user@2001:da8:215:c252:19b6:f7da:70f4:e333] has quit [Ping timeout: 245 seconds] | 18:48 | |
-!- HeikoS [~heiko@f052017018.adsl.alicedsl.de] has joined #shogun | 18:50 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:50 | |
thoralf | Hallo Heiko. | 18:51 |
thoralf | Falls Du kurz Zeit hast: https://github.com/shogun-toolbox/shogun/pull/1243 | 18:52 |
thoralf | iglesiasg said, he's no Makefile expert. But I claim you are. ;) | 18:53 |
@HeikoS | thoralf: Hi! | 19:03 |
@HeikoS | thoralf: I just looked at it, your claim is only 50% true though | 19:05 |
@HeikoS | looks good to me | 19:05 |
@HeikoS | but Id like to wait for others to confirm that ;) | 19:05 |
@HeikoS | thoralf: checking out the pythonevn stuff, pretty awesome | 19:05 |
thoralf | Yeah. | 19:05 |
@HeikoS | wiking: around=? | 19:13 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 19:20 | |
shogun-notifier- | shogun: van51 :develop * 14cbb3c / / (3 files): https://github.com/shogun-toolbox/shogun/commit/14cbb3c564a8f72b67d41929881d404e924b00dc | 19:20 |
shogun-notifier- | shogun: HashedSparseFeatures class | 19:20 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 37b3940 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/37b39407fbd2d51e0080ffa0e5bc0b3e260a7937 | 19:20 |
shogun-notifier- | shogun: Merge pull request #1245 from van51/feature/hashing | 19:20 |
shogun-notifier- | shogun: | 19:20 |
shogun-notifier- | shogun: HashedSparseFeatures class | 19:20 |
-!- hushell [~hushell@c-24-21-220-10.hsd1.or.comcast.net] has joined #shogun | 19:21 | |
@HeikoS | lambday: currently looking into the clone bug you found | 19:30 |
@HeikoS | lambday: thanks for reporting! | 19:30 |
shogun-buildbot | build #1254 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1254 blamelist: van51 <vangelis_51@hotmail.com> | 19:31 |
lambday | HeikoS: hey! | 19:33 |
@HeikoS | hi | 19:34 |
lambday | HeikoS: I changed the template thing | 19:34 |
lambday | HeikoS: now will have to make sure that the unit-tests work properly and then I'll send PR | 19:34 |
@HeikoS | ok | 19:34 |
@HeikoS | sounds good | 19:34 |
lambday | HeikoS: amazing about the engine and job :D | 19:35 |
lambday | HeikoS: I am in :D | 19:35 |
@HeikoS | lambday: good to hear :) we would be very happy if you stayed with us after gsoc | 19:35 |
@HeikoS | lambday: any reasons why you did not initialise the m_operator in SparseMatrixOperator? | 19:35 |
lambday | HeikoS: wait... I am checking | 19:36 |
@HeikoS | lambday: I think this is one source of the bug | 19:36 |
@HeikoS | lambday: dont worry, I will investigate a bit nowq | 19:37 |
lambday | HeikoS: alright... I'll check it too | 19:37 |
lambday | so, in init I should do something like m_operator=SGSparseMatrix<T>(); | 19:38 |
@HeikoS | lambday: I think this is any automatically isnt it? | 19:39 |
lambday | HeikoS: yes that's why I didn't do anythng explicitely :-/ | 19:40 |
@HeikoS | lambday: you are right with that, my fault, checking other things now | 19:40 |
lambday | HeikoS: I am not sure what do you mean by initializing this one :( | 19:40 |
lambday | HeikoS: okay | 19:40 |
@HeikoS | lambday: found the bug, its clone | 19:41 |
shogun-buildbot | build #1050 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1050 | 19:42 |
lambday | HeikoS: what is it? | 19:42 |
shogun-buildbot | build #1051 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1051 blamelist: Soeren Sonnenburg <sonne@debian.org> | 19:43 |
@HeikoS | some NULL arrays are not copied, but ignored so they stay uninitialised | 19:43 |
shogun-buildbot | build #1255 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1255 blamelist: Soeren Sonnenburg <sonne@debian.org> | 19:43 |
lambday | okay.. I'll add a unit test for this when you fix it | 19:46 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 19:47 | |
-!- travis-ci [~travis-ci@ec2-107-22-152-223.compute-1.amazonaws.com] has joined #shogun | 19:48 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/9104709 | 19:48 |
-!- travis-ci [~travis-ci@ec2-107-22-152-223.compute-1.amazonaws.com] has left #shogun [] | 19:48 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 19:50 | |
@HeikoS | lambday: still around? | 19:53 |
@HeikoS | lambday: so you registered sparse vectors as CT_SCALAR and ST_SPARSE | 19:53 |
-!- lisitsyn [~lisitsyn@92-240-133-94.clients.tlt.100megabit.ru] has joined #shogun | 19:53 | |
lisitsyn | ohh | 19:53 |
@HeikoS | but sparse matrices as CT_MATRIX and CT_SPARSE | 19:54 |
@HeikoS | lisitsyn: hey! | 19:54 |
lisitsyn | HeikoS: hey | 19:54 |
@HeikoS | lisitsyn: where are you? :) | 19:54 |
lisitsyn | HeikoS: just got back home | 19:54 |
lisitsyn | I am like pheew | 19:54 |
@HeikoS | lisitsyn: good trip? | 19:54 |
lisitsyn | HeikoS: the issue is that I feel a bit pheeew about russia now | 19:55 |
@HeikoS | lisitsyn: ah, Berlin got you? :D | 19:55 |
lisitsyn | HeikoS: yes totally | 19:55 |
@HeikoS | lisitsyn: many Russians here, you should come ;) | 19:55 |
pickle27 | from what i' | 19:55 |
lisitsyn | HeikoS: I am afraid I could do that one day | 19:55 |
@HeikoS | still enjoying the city, will depart tomorrow | 19:55 |
lisitsyn | HeikoS: have you been to einstein stammhaus? | 19:56 |
pickle27 | sorry from what I've heard Berlin would make a lot of places seem pheew | 19:56 |
@HeikoS | lisitsyn: not yet | 19:56 |
@HeikoS | pickle27: yeh its kinda nice here | 19:56 |
lisitsyn | HeikoS: you should try their apfelstrudel and espresso I think! | 19:56 |
@HeikoS | the only thing is: if you are here, you tend to not work but do other nice things ;) | 19:56 |
@HeikoS | lisitsyn: haha :) apfelstrudel and espresso! :) | 19:56 |
lisitsyn | not really cheap (these two ones would cost like around 10EUR) | 19:57 |
lisitsyn | pickle27: well as I have not been to any foreign country before | 19:57 |
pickle27 | ah | 19:58 |
lisitsyn | ;) | 19:58 |
@HeikoS | lisitsyn: really glad you liked it! :) | 19:58 |
pickle27 | travelling is fun! | 19:58 |
lisitsyn | pickle27: well I started to earn some enough money so I get it now ;) | 19:59 |
@HeikoS | lambday: something is weird in the Parameter::add things for example | 19:59 |
@HeikoS | Parameter::add_matrix(SGSparseVector<int32_t>** param, | 19:59 |
@HeikoS | index_t* length_y, index_t* length_x, | 19:59 |
@HeikoS | const char* name, const char* description) { | 19:59 |
@HeikoS | TSGDataType type(CT_MATRIX, ST_SPARSE, PT_INT32, length_y, | 19:59 |
lisitsyn | HeikoS: I should be back and check some other cities too! | 19:59 |
pickle27 | awesome! | 19:59 |
@sonney2k | lisitsyn, what did you do on sunday? | 19:59 |
@HeikoS | lisitsyn: you should, Berlin is among the nicest though! | 19:59 |
lisitsyn | sonney2k: hauptbahnhof shopping :D | 19:59 |
@sonney2k | pfff | 19:59 |
lisitsyn | well at least I have found something for my mother (she has birthday in a few days) | 20:00 |
lisitsyn | HeikoS: oh man I want to get back already | 20:00 |
lambday | HeikoS: add matrix? | 20:00 |
lambday | HeikoS: I didn't add that | 20:01 |
@HeikoS | lambday: ah I see | 20:01 |
lambday | oh | 20:01 |
lambday | nono | 20:01 |
@HeikoS | lambday: even so, is that correct, see email | 20:01 |
@HeikoS | lisitsyn: hehe | 20:01 |
@HeikoS | lisitsyn: you should come back, and next time hang out with us some more :) | 20:01 |
gsomix | lisitsyn, hi! | 20:01 |
lambday | HeikoS: checking | 20:02 |
lambday | HeikoS: | 20:05 |
@HeikoS | lambday: ah found some more bugs in clone :( | 20:05 |
lambday | Parameter::add(SGSparseVector<complex64_t> has all CT_SCALARs | 20:05 |
@HeikoS | lambday: I know | 20:05 |
@HeikoS | sonney2k: how are sparse vectors represented | 20:06 |
@HeikoS | CT_SCALAR ot CT_VECTOR? | 20:06 |
lambday | I mean Parameter::add(SGSparseVector<T> has all CT_SCALARs | 20:06 |
@HeikoS | currenlty confused | 20:06 |
@sonney2k | HeikoS, scalars yes | 20:06 |
@HeikoS | sonney how about sparse matrix? | 20:06 |
@HeikoS | sonney2k: and when do we use CT_VECTOR with ST_SPARSE? | 20:07 |
@sonney2k | HeikoS, dirk designed it that way | 20:07 |
@HeikoS | sonney2k: its fine, just need to know, so CT_VECTOR is never used along with ST_SPARSE ? | 20:08 |
@sonney2k | HeikoS, sparse matrix would be CT_SPARSE IMHO | 20:08 |
@HeikoS | CT_SPARSE? | 20:08 |
@HeikoS | we only have scalar, vector, matrix | 20:08 |
@sonney2k | well we don't have that yet | 20:08 |
@HeikoS | and ST_NONE and ST_SPARSE | 20:08 |
@HeikoS | sonney2k: well yes there are combinations | 20:09 |
@sonney2k | I am not up-to-date | 20:09 |
@sonney2k | the idea was that one could have vectors/strings in a matrix | 20:09 |
@HeikoS | sonney I see, makes sense | 20:09 |
@HeikoS | sonney2k: but how about sparse vectors | 20:10 |
@sonney2k | IMHO it makes things too complex | 20:10 |
@HeikoS | so first thing is: sparse vectors of scalars | 20:10 |
@sonney2k | we don't need that | 20:10 |
@HeikoS | this would be CT_SCALAR and ST_SPARSE | 20:10 |
@HeikoS | and then sparse matrices, how do we do that: in the code we have: ST_SPARSE and CT_MATRIX | 20:10 |
@HeikoS | but thats weird | 20:10 |
@HeikoS | not consistent | 20:10 |
@sonney2k | HeikoS, well we don't have matrices of strings either | 20:12 |
@HeikoS | sonney2k: no its all just scalar | 20:12 |
@HeikoS | sonney2k: how do we represent sparse matrices of floats? | 20:12 |
thoralf | HeikoS: Lists of sparse vectors ;) | 20:13 |
@sonney2k | well I would be totally fine if we just had a container type scalar/vector/...sparse_matrix | 20:14 |
@sonney2k | with some primitive type | 20:14 |
@HeikoS | sonney2k: I dont want to change things, just understand | 20:14 |
@sonney2k | and rather sgstring as container type | 20:14 |
@sonney2k | I am saying that this is how I would have liked it to be | 20:15 |
@HeikoS | sonney2k: ok I agree on that, but for now I want to fix that bug | 20:15 |
@sonney2k | now we have for strings IIRC CT_VECTOR with ST_STRING | 20:15 |
@HeikoS | and its related to sparse matrix things | 20:15 |
@sonney2k | so that is why ST_SPARSE is the correct thing to do with this framework | 20:16 |
@HeikoS | sonney2k: I am just wondering how to represent sparse matrices correct | 20:16 |
@sonney2k | so you have a vector of sparse | 20:16 |
@HeikoS | yes, thats CT_SCALAR and ST_SPARSE | 20:17 |
@sonney2k | yeah | 20:17 |
@HeikoS | and now I want sparse matrix, which seems to be CT_MATRIX and ST_SPARSE | 20:17 |
thoralf | I like the scipy compressed sparse matrices. | 20:17 |
@sonney2k | no and no | 20:17 |
@sonney2k | shogun sparse vector === CT_SCALAR + ST_SPARSE | 20:18 |
@HeikoS | sonney2k: I found this: | 20:18 |
@sonney2k | shogun sparse matrix == CT_VECTOR + ST_SPARSE | 20:18 |
@HeikoS | void Parameter::add(SGSparseMatrix<bool>* param, | 20:18 |
@HeikoS | const char* name, const char* description) | 20:18 |
@HeikoS | { | 20:18 |
@HeikoS | TSGDataType type(CT_SGMATRIX, ST_SPARSE, PT_BOOL, ¶m->num_vectors, | 20:18 |
@HeikoS | ¶m->num_features); | 20:18 |
@HeikoS | add_type(&type, ¶m->sparse_matrix, name, description); | 20:18 |
@HeikoS | } | 20:18 |
@sonney2k | thoralf, heh they have 3 formats - so which one? | 20:19 |
@HeikoS | whats that? | 20:19 |
thoralf | sonney2k: compressed row/columns | 20:19 |
@sonney2k | LIL, CSC, CRS | 20:19 |
@HeikoS | sonney2k: I see, whats this code snippet doing the? | 20:19 |
@sonney2k | thoralf, they are static though... | 20:19 |
@HeikoS | there | 20:19 |
@HeikoS | sonney2k: any ideas? | 20:19 |
@sonney2k | thoralf, I mean no way to add a vector / insert | 20:20 |
thoralf | sonney2k: Yeah, that's right. | 20:21 |
@sonney2k | HeikoS, ohh I see an issue - with CT_VECTOR you cannot say how many dims the sparse matrix has | 20:21 |
@HeikoS | sonney2k: so thats might be the reason why it is CT_MATRIX? | 20:22 |
thoralf | sonney2k: Nobody wants to change sparse matrices... I always treat them as if they were read-only ;) | 20:22 |
@HeikoS | sonney2k: that makes sense | 20:22 |
thoralf | sonney2k: Btw: Appending vectors works well. | 20:22 |
@sonney2k | thoralf, I mean for dense matrix types it totally makes sense to use just one big array (lapack & friends are optimized for that like hell) no idea if sparse libs have similiar optimiaztions | 20:22 |
thoralf | sonney2k: At least for CSR | 20:22 |
@sonney2k | HeikoS, yeah I guess but it is totally crap | 20:23 |
@HeikoS | sonney2k: how would be the best way? | 20:23 |
@HeikoS | just have container types only I guess: SCALAR, VECTOR, MATRIX, SPARSE_VECTOR, SPARSE_MATRIX? | 20:23 |
@sonney2k | HeikoS, doing it with CT_SGSPARSEMATRIX | 20:23 |
@HeikoS | and drop ST? | 20:24 |
@sonney2k | and CT_SGSPARSEVECTOR | 20:24 |
@sonney2k | yeah I would love to drop ST_* | 20:24 |
@HeikoS | how to do strings then? | 20:24 |
@HeikoS | vectors of chars? | 20:24 |
@sonney2k | but I guess we can only drop it once strings are gone | 20:24 |
thoralf | sonney2k: Hmm, I think sparse matrices are hard to optimize. I'm just thinking of dot products. | 20:25 |
thoralf | sonney2k: So many conditions. I'd like to know how sparse a vector needs to be for that to pay. | 20:25 |
@sonney2k | strings would be CT_SGSTRINGLIST | 20:25 |
@HeikoS | I see, so no strings of matrices etc? | 20:25 |
@sonney2k | thoralf, I guess people like the one doing gotblas would do that | 20:25 |
@HeikoS | only SGObject | 20:25 |
@HeikoS | sonney2k: man, this should have been done before clone/equals | 20:26 |
thoralf | gotblas | 20:26 |
thoralf | gotblas? | 20:26 |
@sonney2k | HeikoS, I admittedly would prefer doing a new type for that when it happens | 20:26 |
@HeikoS | soooo much refactoring now | 20:26 |
@sonney2k | thoralf, gotoblas | 20:26 |
thoralf | Oh, damn. Gotta got. See you! | 20:26 |
@HeikoS | thoralf: bye! | 20:26 |
@sonney2k | thoralf, cu!! | 20:26 |
@HeikoS | sonney2k: parameter framework could be redone somehow then | 20:28 |
@HeikoS | most classes dont interact that much | 20:28 |
@sonney2k | I suspect it woudl be much easier this way | 20:29 |
lambday | HeikoS: I'll be back later | 20:30 |
lambday | see you | 20:30 |
@HeikoS | sonney2k: mmh, so we would have to rewritte: adding/storing/registering and then load/save, and migrate(useless anyway in its current state) | 20:30 |
@HeikoS | lambday: bye! | 20:30 |
-!- lambday [67157e4f@gateway/web/freenode/ip.103.21.126.79] has quit [] | 20:30 | |
@HeikoS | sonney2k: equals/clone once more | 20:31 |
@sonney2k | HeikoS, sorry is it really that much? | 20:31 |
@HeikoS | sonney2k: dont know | 20:31 |
@HeikoS | trying to find out :) | 20:31 |
@sonney2k | HeikoS, I mean ST_* exisits only for NONE and STRING | 20:31 |
@sonney2k | so it cannot be | 20:31 |
@sonney2k | I think just string needs to be done differently | 20:31 |
@sonney2k | then all good | 20:31 |
@HeikoS | sonney2k: save/load/migrate/equals/clone are all totally dependent on the structure CT/ST | 20:32 |
@HeikoS | if we drop one, these basically have to be re-done | 20:32 |
@HeikoS | and when we do this, we can also change the TSGDataType | 20:32 |
@HeikoS | because its totally annoying | 20:32 |
@HeikoS | and then we can also do things in a way that migration is easier | 20:32 |
@sonney2k | sure just drop this parameter but then all the rest will still work | 20:32 |
@HeikoS | (allocating paramters from scratch etc) | 20:32 |
@HeikoS | sonney2k: dont know whether all that is feasible for someone doing other things .... | 20:33 |
@HeikoS | might be a good GSoC project | 20:33 |
@HeikoS | but ultra hard | 20:33 |
@sonney2k | HeikoS, you are the expert with migration / modsel / parameters | 20:34 |
@sonney2k | who would know if not you | 20:34 |
@HeikoS | sonney2k: you :D | 20:34 |
@sonney2k | but just removing SG_* seems very easy! | 20:34 |
@sonney2k | ST_* | 20:34 |
@HeikoS | sonney2k: ok, maybe thats a good start | 20:34 |
@sonney2k | simply stuff later is a different problem | 20:35 |
@HeikoS | sonney2k: what do you think about all those types that are blowing up Parameter.h? | 20:35 |
@HeikoS | uint16_t? | 20:36 |
@HeikoS | I never used that | 20:36 |
@HeikoS | rather float64, float32, int32, int64, char | 20:36 |
@HeikoS | not even unsigned | 20:36 |
@HeikoS | sonney2k: oh on another thing: | 20:37 |
@HeikoS | sonney2k: the spectrum kernel currently is done via those obtain_from_* method of string features | 20:37 |
@HeikoS | which then generates this sparse representation of the k-mer counts | 20:37 |
@HeikoS | and those are given to a kernel | 20:37 |
@HeikoS | sonney2k: so no model-selection of k-mer length | 20:38 |
@HeikoS | since one cannot do model-selection over features | 20:38 |
@sonney2k | HeikoS, I am only using unsigned stuff like uint16_t / 32_t 64_t for strings | 20:38 |
@sonney2k | HeikoS, I did some other spectrum features using all k=1...max_K | 20:39 |
@sonney2k | with weights | 20:39 |
@sonney2k | forgot what the name was | 20:40 |
@sonney2k | but maybe it has the same problem | 20:40 |
@HeikoS | sonney2k: can the kernel select the weights? | 20:40 |
@sonney2k | don't remember | 20:40 |
@HeikoS | ok | 20:40 |
@sonney2k | IMHO it is not that much of a problem | 20:40 |
@HeikoS | sonney2k: no problem, but no automated model selection | 20:41 |
lisitsyn | pickle27: hey I just checked andreas mail | 20:41 |
@HeikoS | sonney2k: btw model selection framework might be a bit useless anyway | 20:42 |
pickle27 | lisitsyn: yeah I've been working on this today | 20:42 |
@HeikoS | sonney2k: what counts more is cross-validation | 20:42 |
@HeikoS | since grid-search is just a loop | 20:42 |
pickle27 | lisitsyn: Im actually about to send an reply | 20:42 |
@HeikoS | which users could easily set up themselved | 20:42 |
@HeikoS | s | 20:42 |
lisitsyn | pickle27: okay sure! | 20:42 |
@HeikoS | sonney2k: but is very hard to maintain | 20:42 |
@HeikoS | if we offer it | 20:42 |
lisitsyn | pickle27: I am on vacations still so I'd try to do my best to get into algorithms and help with further planning | 20:43 |
@HeikoS | sonney2k: at least for grid-search | 20:43 |
@HeikoS | sonney2k: gradient based things are different | 20:43 |
pickle27 | lisitsyn: yeah no prob, I have still got a bunch of things I want to get done before getting fully into example mode | 20:43 |
lisitsyn | pickle27: is everything alright with your masters thesis? | 20:44 |
pickle27 | lisitsyn: yeah I submitted and my defence is on Aug 1! | 20:44 |
lisitsyn | oh I see | 20:44 |
pickle27 | I meant shogun things to get done! | 20:44 |
lisitsyn | what do you mean? by aug 1? | 20:45 |
pickle27 | I mean there is a few more techniques I want to finish coding for shogun before I start putting all my attention to the final examples | 20:45 |
pickle27 | Aug 1 is when I present my thesis | 20:46 |
lisitsyn | ahh alright | 20:46 |
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has joined #shogun | 20:51 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 20:52 | |
@iglesiasg | hi all | 20:52 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 248 seconds] | 20:52 | |
@sonney2k | hey iglesiasg - back home? | 21:01 |
@iglesiasg | sonney2k: hi! yeah, I got here yesterday at night | 21:01 |
@iglesiasg | sonney2k: are you already in holiday? | 21:01 |
@HeikoS | iglesiasg: hi! | 21:02 |
@iglesiasg | HeikoS: thanks for setting up the doodle! | 21:02 |
@sonney2k | iglesiasg, yeah watching the sun set over the sea currently | 21:02 |
@HeikoS | iglesiasg: yes, good to have the meeting soon | 21:02 |
@HeikoS | sonney2k: nice :D | 21:02 |
@iglesiasg | sonney2k: it sounds nice | 21:03 |
@sonney2k | doing (late) nips reviews | 21:03 |
@sonney2k | all rejects IMHO :/ | 21:03 |
@iglesiasg | haha | 21:04 |
@sonney2k | HeikoS, never get yourself's paper reviewed by me | 21:04 |
@sonney2k | I don't remember accepting papers | 21:04 |
@iglesiasg | we better hope for our papers not to be reviewed by you :P | 21:04 |
@HeikoS | sonney2k: maybe you have mine :) | 21:04 |
@HeikoS | kameleon | 21:05 |
@iglesiasg | HeikoS: did you submit sth to NIPS? | 21:05 |
@HeikoS | iglesiasg: yes | 21:05 |
@HeikoS | some kernel mcmc stuff | 21:05 |
@iglesiasg | HeikoS: interesting, I wish you luck with it! | 21:05 |
@sonney2k | iglesiasg, well we all have seen HeikoS talks so we can be pretty sure that this gets accepted | 21:08 |
@HeikoS | iglesiasg: thanks! | 21:08 |
@HeikoS | sonney2k: writing is different ;) | 21:08 |
@HeikoS | but I think it will be fine | 21:08 |
@iglesiasg | sonney2k: btw, I was gladly surprised with Berlin. I found it an amazing place | 21:14 |
@sonney2k | iglesiasg, yeah I like it too :) | 21:15 |
@HeikoS | it is amazing :) I would find it hard to work here though - so many other nice things to do :) | 21:15 |
@HeikoS | going now, see you! | 21:21 |
@iglesiasg | HeikoS, sonney2k: BTW, could any of you take a quick look at thoralf PR when you get some time? | 21:22 |
@HeikoS | iglesiasg: I did, I think its ok, but I dont know enough of makefiles t omerge it, sonney2k, wiking, or lisitsyn should have a look | 21:23 |
@iglesiasg | HeikoS: all right | 21:23 |
@iglesiasg | we are in the same boat about Makefiles hehe | 21:23 |
@iglesiasg | bye HeikoS | 21:24 |
@sonney2k | iglesiasg, yeah had a look - not in mergeable state yet | 21:25 |
@iglesiasg | sonney2k: because of the crash in travis build? | 21:26 |
@iglesiasg | I think that may actually be my fault, it seems to be something is CustomMahalanobisDistance | 21:26 |
@sonney2k | iglesiasg, no not because of that | 21:29 |
-!- HeikoS [~heiko@f052017018.adsl.alicedsl.de] has quit [Quit: Leaving.] | 21:53 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 22:20 | |
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has joined #shogun | 22:31 | |
gsomix | refactoring, refactoring, refactoring | 22:43 |
@sonney2k | iglesiasg, wrt the PR I have to discuss a couple of things ... | 22:59 |
@iglesiasg | sonney2k: the LMNN PR, right? Tell me about it. | 23:01 |
@iglesiasg | sonney2k: after Saturday's conversation I realized there is no need to make the methods static to avoid including LMNNImpl.h in LMNN.h, just need to do this forward declaration of the class CLMNNImpl. So I was thinking of doing that and make the methods non-static | 23:03 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 23:30 | |
--- Log closed Tue Jul 16 00:00:33 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!