--- Log opened Thu Jun 20 00:00:53 2013 | ||
-!- zxtx [~zv@rrcs-74-62-200-195.west.biz.rr.com] has quit [Ping timeout: 260 seconds] | 00:02 | |
-!- zxtx [~zv@rrcs-74-62-200-195.west.biz.rr.com] has joined #shogun | 00:12 | |
-!- zxtx [~zv@rrcs-74-62-200-195.west.biz.rr.com] has quit [Ping timeout: 240 seconds] | 00:53 | |
-!- zxtx [~zv@rrcs-74-62-200-195.west.biz.rr.com] has joined #shogun | 01:00 | |
-!- nube [~rho@49.244.82.20] has joined #shogun | 02:10 | |
-!- nube [~rho@49.244.82.20] has quit [Quit: Leaving.] | 02:15 | |
-!- nube [~rho@49.244.82.20] has joined #shogun | 02:15 | |
-!- nube [~rho@49.244.82.20] has quit [Quit: Leaving.] | 03:53 | |
-!- foulwall [~foulwall@2001:da8:215:503:78d3:5735:f39d:9bb] has joined #shogun | 04:43 | |
-!- zxtx [~zv@rrcs-74-62-200-195.west.biz.rr.com] has quit [Ping timeout: 248 seconds] | 04:44 | |
-!- foulwall [~foulwall@2001:da8:215:503:78d3:5735:f39d:9bb] has quit [Remote host closed the connection] | 05:03 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 05:48 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 06:19 | |
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 264 seconds] | 06:22 | |
-!- lambday [67157e37@gateway/web/freenode/ip.103.21.126.55] has joined #shogun | 06:57 | |
-!- nube [~rho@116.90.239.3] has joined #shogun | 07:14 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 07:22 | |
-!- lambday [67157e37@gateway/web/freenode/ip.103.21.126.55] has quit [Ping timeout: 250 seconds] | 07:22 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] | 07:36 | |
-!- lambday [67157e37@gateway/web/freenode/ip.103.21.126.55] has joined #shogun | 08:10 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 08:17 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 08:24 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Remote host closed the connection] | 08:35 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] | 09:12 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 09:19 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] | 09:30 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 09:31 | |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:36 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 09:36 | |
-!- stepan_ [4d4b4af8@gateway/web/freenode/ip.77.75.74.248] has joined #shogun | 09:45 | |
stepan_ | Hi, I'm using shogun::CGMM that needs lapack library, clapack.h. After downloading lapack and clapack.h, compiling my code returns error: error: declaration of C function ‘int dposv_(const char*, const int*, const int*, double*, const int*, double*, const int*, int*)’ conflicts with ../CLAPACK-3.2.1/INCLUDE/clapack.h:3314: error: previous declaration ‘int dposv_(char*, integer*, integer*, doublereal*, integer*, doublereal*, inte | 09:49 |
---|---|---|
stepan_ | Please how could I use shogun with lapack? | 09:49 |
@iglesiasg | stepan_: what OS are you using? | 09:53 |
stepan_ | linux 64bit, Ubuntu | 09:53 |
@iglesiasg | stepan_: then it is maybe easier to install lapack via apt-get | 09:55 |
@iglesiasg | apt-get install liblapack-dev I think | 09:55 |
stepan_ | yes, but debian lapack package does not have header files...exactly the package you mentioned | 09:56 |
@iglesiasg | you mean the package doesn't install the header? | 09:57 |
@iglesiasg | are you sure? I think it is enough to use apt-get for lapack with Shogun, no need of manual installing | 09:57 |
votjakovr | stepan_: which version of ubuntu are you using? | 09:57 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 09:58 | |
-!- mode/#shogun [+o lisitsyn] by ChanServ | 09:58 | |
stepan_ | actually I'm using debian squeeze in virtual environment, sorry for wrong info | 09:59 |
lambday | lisitsyn: доброе утро :) | 10:00 |
votjakovr | stepan_: i'm not shure but, try to check /usr/include/atlas | 10:00 |
@lisitsyn | hey there | 10:01 |
@lisitsyn | lambday: eine wundershoene guten morgen | 10:01 |
votjakovr | stepan_: clapack.h might be there | 10:01 |
stepan_ | votjakovr: atlas clapack.h file is working. Thanks! | 10:02 |
votjakovr | stepan_: ;) | 10:02 |
-!- lambday_ [67157c37@gateway/web/freenode/ip.103.21.124.55] has joined #shogun | 10:02 | |
lambday_ | its /usr/include/clapack.h I think | 10:02 |
lambday_ | lisitsyn: sorry I missed that.. got dc :( | 10:03 |
-!- hushell [~hushell@c-24-21-141-32.hsd1.or.comcast.net] has joined #shogun | 10:03 | |
-!- lambday [67157e37@gateway/web/freenode/ip.103.21.126.55] has quit [Ping timeout: 250 seconds] | 10:04 | |
-!- lambday_ [67157c37@gateway/web/freenode/ip.103.21.124.55] has quit [Client Quit] | 10:05 | |
-!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has joined #shogun | 10:05 | |
-!- stepan_ [4d4b4af8@gateway/web/freenode/ip.77.75.74.248] has quit [Quit: Page closed] | 10:08 | |
@lisitsyn | lambday: missed what? ;) | 10:10 |
lambday | lisitsyn: what you said after I said good morning :P log not updated yet :P | 10:11 |
lambday | lisitsyn: by the way, where is the workshop is gonna be held? | 10:12 |
lambday | berlin? | 10:12 |
@iglesiasg | yes | 10:14 |
lambday | iglesiasg: okay :) | 10:14 |
@lisitsyn | lambday: yes sure berlin | 10:30 |
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun | 10:41 | |
@wiking | lisitsyn: r u going to be there on 14th? | 10:44 |
@lisitsyn | wiking: ja | 10:44 |
-!- foulwall [~foulwall@2001:da8:215:c252:6800:d0c6:c4d3:85c5] has joined #shogun | 10:44 | |
@wiking | lisitsyn: ok | 10:44 |
@wiking | lisitsyn: i'm just buying my tix | 10:44 |
@lisitsyn | wiking: I have my flight morning of 15th | 10:44 |
@lisitsyn | wiking: my visa is ready btw | 10:45 |
@wiking | aaah ok | 10:45 |
@wiking | lisitsyn: \o/ | 10:45 |
@lisitsyn | everything went smooth | 10:45 |
@wiking | cool ok so i go back on the 14th | 10:45 |
@lisitsyn | wiking: when are you in the city? | 10:45 |
@lisitsyn | btw where you're going to stay? | 10:45 |
@wiking | mmm i have friends living in berlin | 10:46 |
@wiking | so i might stay with them | 10:46 |
@lisitsyn | I see | 10:46 |
@lisitsyn | wiking: seems you have friends all around the globe ;) | 10:46 |
@wiking | naaah berlin is an exception | 10:46 |
@iglesiasg | lisitsyn, wiking : do you recognize this error in travis build ImportError: cannot import name SerializableJsonFile? | 10:48 |
@iglesiasg | https://travis-ci.org/shogun-toolbox/shogun/jobs/8268716 | 10:48 |
@lisitsyn | iglesiasg: not really | 10:49 |
@wiking | woohooo | 10:49 |
@wiking | never seen this | 10:49 |
@iglesiasg | no idea why it happened | 10:50 |
@iglesiasg | wiking: is it possible to force re-build in travis? | 11:21 |
@iglesiasg | just to see if the error above still happens | 11:21 |
-!- foulwall [~foulwall@2001:da8:215:c252:6800:d0c6:c4d3:85c5] has quit [Remote host closed the connection] | 11:27 | |
@wiking | iglesiasg: no unfortunately i dont know any way to force a rebuild.... only with a dummy commit :P | 11:40 |
van51 | hey guys | 11:41 |
van51 | I have an interview for a master's today and I 'm not sure whether I'll make it for the meeting :( | 11:42 |
van51 | it's at 13 UTC, but I don't know if there will be delays on their schedule or anything | 11:43 |
van51 | also do you know where sonney is? | 11:43 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 11:45 | |
shogun-notifier- | shogun: iglesias :develop * bf430bc / src/shogun/distance/ (6 files): https://github.com/shogun-toolbox/shogun/commit/bf430bc660897b5419fb7a2527e3fab2a964a57e | 11:45 |
shogun-notifier- | shogun: Minor changes in Custom, Dense and base distances. | 11:45 |
shogun-notifier- | shogun: Mainly move a few methods from headers to implementation files. | 11:45 |
shogun-notifier- | shogun: Fernando Iglesias :develop * d0e18e5 / src/shogun/distance/ (6 files): https://github.com/shogun-toolbox/shogun/commit/d0e18e515154d0eeace86fa2f90719f58945dc67 | 11:45 |
shogun-notifier- | shogun: Merge pull request #1178 from iglesias/develop | 11:45 |
shogun-notifier- | shogun: | 11:45 |
shogun-notifier- | shogun: Minor changes in Custom, Dense and base distances. | 11:45 |
@lisitsyn | van51: could you please send me what you would like to say about your project | 11:58 |
@lisitsyn | as some application etc | 11:58 |
shogun-buildbot | build #1122 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1122 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 12:01 |
shogun-buildbot | build #954 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/954 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 12:05 |
shogun-buildbot | build #955 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/955 blamelist: iglesias <fernando.iglesiasg@gmail.com> | 12:06 |
shogun-buildbot | build #1123 of bsd1 - libshogun is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1123 blamelist: iglesias <fernando.iglesiasg@gmail.com> | 12:09 |
-!- travis-ci [~travis-ci@ec2-54-226-23-108.compute-1.amazonaws.com] has joined #shogun | 12:11 | |
travis-ci | [travis-ci] it's Fernando Iglesias'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/8270588 | 12:11 |
-!- travis-ci [~travis-ci@ec2-54-226-23-108.compute-1.amazonaws.com] has left #shogun [] | 12:11 | |
-!- nube [~rho@116.90.239.3] has quit [Quit: Leaving.] | 12:12 | |
@iglesiasg | lisitsyn, wiking : same error happends in bsd1 | 12:13 |
@lisitsyn | iglesiasg: well soeren changed something in json recently | 12:14 |
@iglesiasg | maybe SerializableJsonFile is not accesible in python in bsd1 longer? | 12:14 |
@iglesiasg | lisitsyn: aham I see | 12:14 |
van51 | lisitsyn: well we haven't settled down on a specific application yet | 12:15 |
van51 | lisitsyn: what I know for sure atm is that it will be a classification task and it will probably be an interactive application | 12:15 |
van51 | lisitsyn: like the user will upload something, a document, a picture w/e and it will be classified accordingly | 12:15 |
shogun-buildbot | build #1246 of deb3 - modular_interfaces is complete: Failure [failed compile python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1246 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 12:16 |
van51 | lisitsyn: the mentors (Olivier and Benoit) suggested a language detection app or a spam filter to show that that the model remains robust | 12:16 |
van51 | even when the feature dimension is reduced | 12:17 |
shogun-buildbot | build #1247 of deb3 - modular_interfaces is complete: Failure [failed compile python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1247 blamelist: iglesias <fernando.iglesiasg@gmail.com> | 12:17 |
van51 | also it's not clear to me yet if we will have to do a seperate app for the fast food project | 12:17 |
@lisitsyn | van51: alright I see | 12:21 |
van51 | lisitsyn: all we have specified for now are the milestones for the project | 12:22 |
van51 | lisitsyn: btw this application has to be ready by the end of GSoC? | 12:22 |
@lisitsyn | van51: yes | 12:23 |
van51 | lisitsyn: or does it have to be semi-completed by the middle-term evaluation? | 12:23 |
@lisitsyn | van51: would be nice to have a prototype by midterm indeed | 12:23 |
van51 | lisitsyn: ok thanks | 12:24 |
van51 | well we just have to agree on a domain then | 12:24 |
van51 | because personally I find text classification a bit boring :P | 12:24 |
@lisitsyn | van51: then we should select something you are not bored with! | 12:24 |
van51 | lisitsyn: yea! | 12:25 |
van51 | lisitsyn: soeren had a nice idea about face detection | 12:25 |
van51 | I will try to subliminally force that :) | 12:26 |
@iglesiasg | I killed shogun :O | 12:27 |
van51 | iglesiasg: should we stop making plans then? | 12:28 |
@lisitsyn | van51: well I am not sure you will beat V-J :) | 12:28 |
van51 | lisitsyn: I just looked V-J up :$ | 12:30 |
van51 | lisitsyn: I don't have much experience in image mining so that's why I found it most interesting | 12:30 |
@lisitsyn | van51: viola-jones | 12:31 |
@lisitsyn | adaboost + haar features | 12:31 |
@iglesiasg | van51: haha yeah | 12:32 |
van51 | lisitsyn: I see | 12:33 |
van51 | lisitsyn: well I 'll discuss it with them and we'll see | 12:33 |
van51 | lisitsyn: maybe it's ok if we don't beat it :P | 12:33 |
@lisitsyn | van51: it is ok as what you do is large-scale or so | 12:34 |
van51 | lisitsyn: does V-J scale well? | 12:35 |
@iglesiasg | van51: from when I implemented V-J I would say a cascaded detector scales well | 12:39 |
@iglesiasg | with the number of training examples I am talking about | 12:40 |
van51 | iglesiasg: ok, thanks :) | 12:45 |
@iglesiasg | it seems it sort of screwed up when I moved a templated method from CustomDistance.h to CustomDistance.cpp | 12:45 |
@iglesiasg | although I don't get why it compiles fine in my machine | 12:47 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 13:06 | |
@wiking | lisitsyn: http://www.liveleak.com/view?i=95c_1371654375#WqxsriSqV6TCxpli.01 | 13:24 |
@wiking | lisitsyn: ROTFL | 13:24 |
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has quit [Quit: Leaving.] | 13:28 | |
@iglesiasg | muahahaha | 13:32 |
lambday | hahahaha :D | 13:36 |
votjakovr | lol :D | 13:36 |
lambday | epic! | 13:37 |
votjakovr | indeed | 13:37 |
@lisitsyn | wiking: hhehe yes that's dangerous | 13:58 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Ping timeout: 255 seconds] | 13:58 | |
@lisitsyn | wiking: that's actually from movie 43 IIRC | 14:03 |
lambday | http://www.youtube.com/watch?v=E3418SeWZfQ | 14:21 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 14:21 | |
@lisitsyn | foulwall: hey there! | 14:22 |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 268 seconds] | 14:22 | |
foulwall | hey lisitsyn | 14:25 |
foulwall | long time no see | 14:25 |
@lisitsyn | foulwall: yeah, so what's up, are you done with your exams? | 14:26 |
-!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has quit [Ping timeout: 250 seconds] | 14:27 | |
foulwall | lisitsyn: yeah | 14:28 |
foulwall | and ~~ full time job from now | 14:29 |
@lisitsyn | foulwall: gsoc is ~ full-time job I hope? :) | 14:29 |
foulwall | yeah | 14:30 |
@lisitsyn | I see | 14:32 |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 14:43 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 14:43 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 14:45 | |
-!- lambday [67157d37@gateway/web/freenode/ip.103.21.125.55] has joined #shogun | 15:01 | |
lambday | the meeting starts in an hour, right? | 15:02 |
@lisitsyn | lambday: yes | 15:03 |
lambday | lisitsyn: soeren won't be here? :( | 15:03 |
@lisitsyn | lambday: no, I'll lead this one | 15:03 |
lambday | lisitsyn: alright :) | 15:03 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 15:04 | |
-!- az_de [82954e22@gateway/web/freenode/ip.130.149.78.34] has joined #shogun | 15:37 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 15:47 | |
shogun-notifier- | shogun: lambday :develop * cacb172 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/cacb172b79d832629472226492c592a3f9c5961d | 15:47 |
shogun-notifier- | shogun: few base classes and implementations for the log-det framework added | 15:47 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 4cf83b1 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/4cf83b1dd301cc67a9b85ba838128e0e2854dd73 | 15:47 |
shogun-notifier- | shogun: Merge pull request #1179 from lambday/feature/log_determinant | 15:47 |
shogun-notifier- | shogun: | 15:47 |
shogun-notifier- | shogun: few base classes and implementations for the log-det framework added | 15:47 |
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun | 15:50 | |
shogun-buildbot | build #956 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/956 blamelist: lambday <heavensdevil6909@gmail.com> | 15:53 |
-!- patrickp [~patrickp@84-75-165-165.dclient.hispeed.ch] has joined #shogun | 15:58 | |
@lisitsyn | lets start in a few minutes! | 15:58 |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: Lost terminal] | 15:59 | |
shogun-buildbot | build #1124 of bsd1 - libshogun is complete: Failure [failed test test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1124 blamelist: lambday <heavensdevil6909@gmail.com> | 16:00 |
lambday | argh! I don't have any idea why these tests work fine locally and never work on bsd :( | 16:01 |
-!- zeller [~zeller@embln.embl.de] has joined #shogun | 16:01 | |
zeller | hello everyone | 16:02 |
@lisitsyn | so everyone is here? | 16:02 |
@lisitsyn | zeller: hey there | 16:02 |
az_de | Hi all! | 16:02 |
pickle27 | hello! | 16:02 |
van51 | hello | 16:02 |
@lisitsyn | hello everyone, welcome to our first GSoC meeting! | 16:02 |
votjakovr | Hi everyone! | 16:02 |
foulwall | hello | 16:02 |
lambday | hello | 16:02 |
@lisitsyn | so the agenda is rather simple | 16:03 |
@lisitsyn | Soeren is busy now so I have to lead this meeting :) | 16:03 |
@lisitsyn | first of all I want to remind you that once GSoC started all students have to send weekly reports | 16:04 |
@lisitsyn | I hope you do remember it is crucial and is a hard requirement | 16:05 |
@lisitsyn | so please don't miss | 16:05 |
-!- gsomix [~gsomix@80.234.20.89] has quit [Ping timeout: 240 seconds] | 16:05 | |
shogun-buildbot | build #1248 of deb3 - modular_interfaces is complete: Failure [failed compile python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1248 blamelist: lambday <heavensdevil6909@gmail.com> | 16:05 |
@lisitsyn | it is just a few lines on your work | 16:05 |
@lisitsyn | so that's clear I guess? :) | 16:05 |
zeller | do they have to officially submit these weekly reports somewhere? | 16:06 |
@lisitsyn | zeller: thanks | 16:06 |
@lisitsyn | reports are supposed to be sent to the mailing list | 16:06 |
@lisitsyn | each monday | 16:06 |
lambday | sending the link of my blog (on which I will update the reports) to the mailing list is cool, right? | 16:07 |
@lisitsyn | lambday: yes indeed but please put some short text to the mail too | 16:07 |
lambday | alright | 16:07 |
@lisitsyn | any other questions with reports? | 16:08 |
@lisitsyn | okay | 16:08 |
@lisitsyn | next point is applications/demos we have to develop this summer | 16:09 |
@lisitsyn | just to recall you, we have decided that each student has to develop a small demo which shows the project he was working at | 16:09 |
shogun-buildbot | build #1125 of bsd1 - libshogun is complete: Failure [failed test test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1125 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 16:09 |
shogun-buildbot | build #1249 of deb3 - modular_interfaces is complete: Failure [failed compile python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1249 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 16:09 |
@lisitsyn | last time we decided to think about possible demos, so lets talk about them one by one | 16:10 |
@lisitsyn | who wants to start? | 16:10 |
pickle27 | I can | 16:10 |
@lisitsyn | az_de: pickle27: lets talk about your project then :) | 16:10 |
@lisitsyn | please continue pickle27 | 16:11 |
pickle27 | I'm going to be working on algorithms for blind source separation | 16:11 |
pickle27 | for an application we're thinking of an website demoing the algorithm on audio and ekg data | 16:11 |
shogun-buildbot | build #957 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/957 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 16:12 |
@lisitsyn | so the user would send some audio file may be | 16:12 |
@lisitsyn | or select one from existing datafiles | 16:12 |
pickle27 | I think that answer will depend on how robust the technqiues are | 16:13 |
az_de | does it have to be a webdemo or just some demo in shogun? | 16:13 |
pickle27 | if they are super robust then definitely let users upload | 16:13 |
-!- travis-ci [~travis-ci@ec2-54-226-23-108.compute-1.amazonaws.com] has joined #shogun | 16:13 | |
travis-ci | [travis-ci] it's Heiko Strathmann'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/8276683 | 16:13 |
-!- travis-ci [~travis-ci@ec2-54-226-23-108.compute-1.amazonaws.com] has left #shogun [] | 16:13 | |
@lisitsyn | az_de: well web based demo is indeed better I think | 16:14 |
pickle27 | az_de, the main advantage would be easy access to the demo | 16:14 |
az_de | ok, if you know how to do that :-) | 16:14 |
pickle27 | possibly getting people interested in actually downloading and using the lib | 16:14 |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has joined #shogun | 16:15 | |
pickle27 | anyways so thats what we're thinking for now we'll continually evaluate as we learn and code more1 | 16:15 |
@lisitsyn | az_de: well I can help there and I hope foulwall will share some experience too :) | 16:15 |
foulwall | hi pickle27 | 16:15 |
pickle27 | hi foulwall | 16:15 |
@lisitsyn | alright so that's the application: separation of audio files (like two songs at the same time) | 16:16 |
@lisitsyn | lets move on to zeller and iglesiasg project :) | 16:16 |
@lisitsyn | iglesiasg: you are on the phone now, right? | 16:17 |
foulwall | hoping you can put your web based demo in https://github.com/shogun-toolbox/shogun-demo | 16:17 |
iglesiasg | yeah! in the subway actually :P | 16:17 |
@lisitsyn | iglesiasg: heh alright | 16:18 |
@lisitsyn | so the project of iglesiasg is large-margin nearest neighbor algorithm | 16:18 |
@lisitsyn | you may know it is a distance learning thing, it learns a transformation of distance (like mahalanobis with custom matrix) | 16:18 |
@lisitsyn | to obtain largest margin in supervised manner | 16:19 |
@lisitsyn | zeller: could you tell us a bit about the applications? | 16:20 |
zeller | the applications I have in mind are mostly from computational biology | 16:20 |
zeller | metagenomics and chemogenomics | 16:20 |
zeller | e.g. to separate samples of microbial DNA depending on the community from which they were sampled | 16:21 |
@lisitsyn | zeller: what makes LMNN special for these fields btw? | 16:21 |
zeller | these communities could be of microbes that colonize the human body | 16:21 |
zeller | often you deal with multiclass problems here sometimes with really many classes | 16:22 |
@lisitsyn | iglesiasg: anything else you'd like to say? ;) | 16:23 |
zeller | that's also the case for the chemogenomics data set: this contains gene expression changes in cell liens treated with ~1000 drugs | 16:23 |
zeller | so you can easily have dozens of classes | 16:23 |
@lisitsyn | zeller: so the application (I think we could go web again) would be some plotting of the dataset or may be graphs | 16:23 |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has quit [Read error: Connection reset by peer] | 16:23 | |
@lisitsyn | that tells us we have improvement over say SVM | 16:24 |
@lisitsyn | which is not that cool for multiclass | 16:24 |
zeller | yes and hopefully we can convince people that it works better than standard PCA :-) | 16:24 |
-!- iglesiasg [~fircuser@2001:6b0:1:1041:1877:6c11:6290:4e60] has joined #shogun | 16:24 | |
-!- gsomix [~gsomix@85.26.234.206] has joined #shogun | 16:24 | |
zeller | that's one point | 16:24 |
pickle27 | amen to that | 16:25 |
@lisitsyn | heh | 16:25 |
pickle27 | PCA is over used lol | 16:25 |
zeller | the other is that these PCA-based techniques are really widespread in these research areas and LMNN can be seen as just a better distance transformation | 16:25 |
@lisitsyn | zeller: yeah so we compare and show an improvement | 16:25 |
zeller | hopefully giving also better results visually | 16:25 |
@lisitsyn | but still fancy web | 16:26 |
zeller | sure | 16:26 |
@lisitsyn | so I hope to get it attractive too | 16:26 |
zeller | although I wouldn't know much about how to set this up... | 16:26 |
@lisitsyn | zeller: okay, anything else you'd like to say? | 16:26 |
@lisitsyn | zeller: no worries | 16:26 |
@lisitsyn | we will guide people here | 16:26 |
zeller | I guess we could leave it here | 16:26 |
pickle27 | I think we're all a little behind on web stuff so we'll figure it out together! | 16:27 |
zeller | sure cool! | 16:27 |
@lisitsyn | alright, thanks zeller | 16:27 |
@lisitsyn | hushell: patrickp: want to continue? | 16:27 |
patrickp | we have actually not yet discussed the demo/applications | 16:27 |
@lisitsyn | patrickp: lets do that now ;) rather short though | 16:28 |
-!- Cheng [3f85cc22@gateway/web/freenode/ip.63.133.204.34] has joined #shogun | 16:28 | |
@lisitsyn | well I know hushell is doing some coarse-to-fine detection, right? | 16:28 |
@lisitsyn | hey Cheng! | 16:28 |
-!- iglesiasg [~fircuser@2001:6b0:1:1041:1877:6c11:6290:4e60] has quit [Ping timeout: 245 seconds] | 16:28 | |
patrickp | there exist a few standard applications of structured output prediction: sequence labeling (pos, ocr) is probably the easiest | 16:28 |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has joined #shogun | 16:28 | |
pickle27 | a good open source ocr demo would be sweet! | 16:29 |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has quit [Client Quit] | 16:29 | |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has joined #shogun | 16:29 | |
foulwall | Hi Cheng! | 16:29 |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has quit [Remote host closed the connection] | 16:29 | |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has joined #shogun | 16:29 | |
Cheng | Hi Sergey, I'm totally messed up with timezones at the moment. At ICML in Atlanta. | 16:29 |
patrickp | hi cheng! | 16:30 |
Cheng | Hi foulwall! | 16:30 |
@lisitsyn | Cheng: I feel your pain :) | 16:30 |
@lisitsyn | australia -> usa | 16:30 |
@lisitsyn | patrickp: alright so anything else? | 16:30 |
Cheng | Sorry to interrupt the discussion | 16:30 |
@lisitsyn | Cheng: I guess you are here to talk a bit about foulwall project right? ;) | 16:30 |
patrickp | Anyway: we could also think of other applications, mostly thinking about computer vision, but they are pretty tricky in this short time frame | 16:30 |
@lisitsyn | patrickp: indeed | 16:31 |
patrickp | hushell: do you have some ideas? | 16:31 |
@lisitsyn | patrickp: NLP thing would be easier I guess? | 16:31 |
iglesiasg | lisitsyn, zeller, thanks for backing me up | 16:31 |
Cheng | lisitsyn: Yes, well to help if I can. | 16:31 |
patrickp | lisitsyn: ocr should be feasible too. just sequence | 16:31 |
@lisitsyn | patrickp: allright good | 16:32 |
patrickp | i'll talk to shell again about this | 16:33 |
@lisitsyn | yes, please think about some concrete app | 16:33 |
patrickp | will do | 16:33 |
@lisitsyn | thanks patrickp! hushell is away I guess :) | 16:33 |
@lisitsyn | Cheng: foulwall: lets talk about your project then | 16:33 |
@lisitsyn | first of all, foulwall is the MVP for demos | 16:34 |
foulwall | lisitsyn: :) | 16:34 |
@lisitsyn | so we are going to use his small framework | 16:34 |
@lisitsyn | to develop these demos | 16:34 |
@lisitsyn | (I hope so!) | 16:34 |
@lisitsyn | other demos are to go still though but that's to be dicussed later | 16:35 |
@lisitsyn | as Cheng is here I'd like to check what we have on mldata integration | 16:35 |
@lisitsyn | Cheng: foulwall can any of you talk on that a bit? | 16:36 |
@lisitsyn | ping ;) | 16:36 |
foulwall | Cheng: Now I've worked out a simple interface on mldata | 16:37 |
Cheng | I have been a bit distracted recently. I sent a list of 6-7 datasets to the shogun list to start of the testing | 16:37 |
Cheng | foulwall: Perhaps you need to summarize what you have done to date. | 16:37 |
foulwall | Cheng: Ok, I'm sorry for my distracted, for the finals... | 16:38 |
@lisitsyn | we have foulwall a bit out of time with his exams so we'd have to get back to the schedule | 16:39 |
foulwall | Cheng, I'll check that email again and make a schedule out | 16:39 |
@lisitsyn | Cheng: could you please tell us what is we are aiming to? | 16:39 |
@lisitsyn | what can we do once foulwall is done with integration | 16:40 |
@lisitsyn | votjakovr: around? | 16:41 |
votjakovr | lisitsyn: yep | 16:41 |
Cheng | The idea is to write a data class in shogun to read (and write) mldata containers. | 16:41 |
@lisitsyn | votjakovr: please prepare to be next ;) | 16:42 |
@lisitsyn | Cheng: so we can just load any mldata dataset in oneliner, right? | 16:42 |
Cheng | Once this is done, interaction with mldata.org's datasets should be much easier. | 16:42 |
votjakovr | lisitsyn: Ok, just let me know when i need to start | 16:43 |
@lisitsyn | I am not sure it makes sense to speak about demos as foulwall's project is all about demos | 16:43 |
@lisitsyn | :) | 16:43 |
@lisitsyn | Cheng: do you have anything to note on the project? | 16:43 |
@lisitsyn | foulwall: and you? | 16:43 |
foulwall | nop | 16:43 |
@lisitsyn | alright | 16:45 |
@lisitsyn | votjakovr: as Heiko is not here you have to talk on your own ;) | 16:45 |
votjakovr | Ok | 16:45 |
votjakovr | As you might know, my GSoC project is about implementing Gaussian Processes for Classification. | 16:45 |
@lisitsyn | Cheng: sorry we have to hurry up a bit ;) | 16:46 |
@lisitsyn | thanks Cheng: | 16:46 |
votjakovr | I had discussed the possible applications of GPC with HeikoS. | 16:46 |
votjakovr | And we decided to make several classification (binary and multiclass) examples with nice real/toy datasets (2d and higher dimensional), which can show an advantages of using Gaussian Processes. | 16:46 |
votjakovr | And also I planned to implement an interactive point-click demo. | 16:46 |
Cheng | nothing to note. | 16:46 |
votjakovr | These examples might be useful for people, who is unfamiliar with GPs. | 16:46 |
votjakovr | And we might get a better results with GPC rather than SVMs in multiclass case :) | 16:47 |
@lisitsyn | votjakovr: indeed if you have a lot of classes | 16:47 |
@lisitsyn | votjakovr: alright so the demo is actually an improvement of the thing foulwall created | 16:48 |
@lisitsyn | http://shogun-toolbox.org/demos/gp/entrance | 16:48 |
votjakovr | lisitsyn: yep, the idea was about implement all features of GPs in one demo | 16:49 |
@lisitsyn | yes that would be nice | 16:49 |
votjakovr | lisitsyn: regression and classification | 16:49 |
@lisitsyn | okay I have not the best understanding of GPs so I am not sure I can comment something ;) | 16:50 |
@lisitsyn | thanks votjakovr | 16:50 |
@lisitsyn | van51: around? | 16:50 |
van51 | lisitsyn: yeap | 16:50 |
van51 | lisitsyn: should I go next? | 16:50 |
@lisitsyn | van51: yes please | 16:51 |
@lisitsyn | van51: I know you didn't come up with anything concrete | 16:51 |
van51 | i will just give a gist of our ideas | 16:51 |
@lisitsyn | yeah sure | 16:51 |
van51 | so, for this GSoC I will implement two different but correlated techniques for learning on large scale data | 16:52 |
van51 | one is based on reducing the feature dimension using hashing | 16:52 |
van51 | the other approximates the feature space ( I think, I haven't looked into that one yet :P ) | 16:52 |
van51 | and as for a demo/application | 16:53 |
van51 | we have been thinking of something web based as well, where the user could upload a piece of text or an image and it would classify it appropriately | 16:53 |
van51 | something like language detection or spam filter | 16:53 |
@lisitsyn | alright! | 16:54 |
@lisitsyn | thanks van51 | 16:54 |
van51 | we haven't decided on the application domain yet, but that is what its infrastacture is going to be | 16:54 |
van51 | np :) | 16:54 |
@lisitsyn | van51: yes, lets decide on the domain | 16:54 |
@lisitsyn | as I said lets choose something you like most | 16:54 |
@lisitsyn | okay so next is gsomix | 16:56 |
gsomix | lisitsyn, ok | 16:56 |
@lisitsyn | gsomix: what can we consider as a demo of your project? | 16:56 |
gsomix | so, this summer I'm working at readers and writers in various file formats. this is technical project. | 16:57 |
gsomix | demos are technical too. | 16:57 |
@lisitsyn | yeah | 16:57 |
@lisitsyn | so my idea would be is to show some graphs | 16:57 |
@lisitsyn | that prove that readers are fast enough | 16:58 |
@lisitsyn | I agree there is no big room for demo-ing | 16:58 |
@lisitsyn | so lets go with some graphs | 16:58 |
@lisitsyn | alright thanks gsomix | 16:59 |
gsomix | This includes: benchmarks; converter from one format to other (like csv->matlab); some real-life demo like load data sets run some machine on it | 16:59 |
gsomix | lisitsyn, yeah, graphs will be nice | 16:59 |
@lisitsyn | yes | 16:59 |
@lisitsyn | loading-saving seems to be a nice application | 17:00 |
@lisitsyn | lambday: here? | 17:00 |
lambday | lisitsyn: yes | 17:00 |
@lisitsyn | lambday: could you please finish then? :) | 17:00 |
lambday | sure :) | 17:00 |
lambday | alright, so, we have this huge Gaussian markov random field.. and the data is real (gathered from an orbiting satelite).. and we are trying to estimate parameters in this... | 17:01 |
lambday | the likelihood expression contains a log determinant term of the precision matrix, which is massive (~10^6) and sparse.. and therefore becomes the bottleneck of estimating parameters..(because of massive memory requirement).. because the usual techniques fail.. | 17:01 |
lambday | my project is about approximating this log determinant of a matrix up to an arbitrary precision which can be computed on a normal computer with, say 4~8 GB ram.. and the approach relies on a number of techniques from numerical linear algebra and complex analysis.. | 17:02 |
lambday | once the project is ready, we have planned to work on this ozone example http://www.ucl.ac.uk/roulette/ozoneexample and estimate parameters.. | 17:02 |
@lisitsyn | lambday: alright and this is going to be the demo | 17:03 |
@lisitsyn | right? | 17:03 |
iglesiasg | it sounds very interesting lambday! | 17:04 |
lambday | lisitsyn: yep | 17:04 |
lambday | iglesiasg: yes :) I'm really excited to see this working | 17:04 |
@lisitsyn | lambday: alright, this is again some graphics so we could go for doing it with d3 too | 17:04 |
@lisitsyn | thanks lambday | 17:05 |
@lisitsyn | alright | 17:05 |
lambday | lisitsyn: we haven't decided anything on that yet.. I'll talk to Heiko regarding that :) | 17:05 |
@lisitsyn | this was long! sorry if I took more time than expected | 17:05 |
az_de | BTW, it is very hot in Berlin: >35°C! | 17:05 |
az_de | .. my brain is melting ... | 17:05 |
@lisitsyn | so, mentors, az_de Cheng patrickp zeller please continue tracking your students :) | 17:06 |
@lisitsyn | az_de: too hot! | 17:06 |
az_de | indeed! | 17:06 |
@lisitsyn | az_de: I hope it won't be that hot when I arrive | 17:06 |
@lisitsyn | students&mentors, please discuss your projects more to make them more concrete | 17:07 |
az_de | no, we are expecting a thunderstorm | 17:07 |
@lisitsyn | and please don't forget about weekly reports | 17:07 |
@lisitsyn | please send them each monday to the mailing list | 17:07 |
@lisitsyn | has anyone got questions? | 17:07 |
lambday | lisitsyn: by monday, you mean as long as its still monday anywhere on earth, we're cool, right? :P | 17:08 |
@lisitsyn | lambday: heh well please send them rather monday morning | 17:08 |
@lisitsyn | or even sunday evening | 17:08 |
lambday | lisitsyn: alright :D haha | 17:09 |
@lisitsyn | as for next meeting | 17:09 |
@lisitsyn | lets schedule it a bit later | 17:09 |
@lisitsyn | let iglesiasg send around a doodle and then we can schedule it | 17:10 |
iglesiasg | sure, I will do this by the end of the weekend | 17:10 |
@lisitsyn | if no questions - thanks for attending it! I hope next time we will all meet | 17:11 |
iglesiasg | let us say so it is a bit before mid term evaluation | 17:11 |
@lisitsyn | as now there is no soeren and heiko that's indeed incomplete :) | 17:11 |
@lisitsyn | iglesiasg: I think a bit before 12th July | 17:12 |
@lisitsyn | before workshop :) | 17:12 |
iglesiasg | lisitsyn: thank you for takinh care of the meeting | 17:12 |
iglesiasg | all right, before workshop | 17:12 |
-!- gsomix [~gsomix@85.26.234.206] has quit [Ping timeout: 255 seconds] | 17:13 | |
@lisitsyn | yeah so lets discuss it later and set up a doodle | 17:13 |
@lisitsyn | thanks everyone | 17:13 |
az_de | Thank you, bye | 17:13 |
lambday | thanks lisitsyn :) | 17:13 |
iglesiasg | thank you people! I will do my best to be in a computer rather than the phone for the next meeting :D | 17:14 |
pickle27 | bye everyone! | 17:15 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 17:15 | |
-!- az_de [82954e22@gateway/web/freenode/ip.130.149.78.34] has left #shogun [] | 17:16 | |
zeller | thansk & bye | 17:17 |
-!- Cheng [3f85cc22@gateway/web/freenode/ip.63.133.204.34] has quit [Quit: Page closed] | 17:18 | |
-!- zeller [~zeller@embln.embl.de] has left #shogun [] | 17:18 | |
-!- patrickp [~patrickp@84-75-165-165.dclient.hispeed.ch] has left #shogun [] | 17:19 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 17:22 | |
iglesiasg | see you people | 17:22 |
-!- iglesiasg [~fircuser@m37-197-66-125.cust.tele2.se] has left #shogun [] | 17:22 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Ping timeout: 246 seconds] | 17:23 | |
lambday | bye guys | 17:24 |
-!- lambday [67157d37@gateway/web/freenode/ip.103.21.125.55] has quit [] | 17:24 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Ping timeout: 276 seconds] | 17:25 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 17:32 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun | 17:32 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 17:37 | |
-!- foulwall_ [~foulwall@li379-21.members.linode.com] has joined #shogun | 17:43 | |
-!- nube [~rho@49.244.56.141] has joined #shogun | 17:49 | |
-!- foulwall_ [~foulwall@li379-21.members.linode.com] has quit [Remote host closed the connection] | 17:49 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 18:02 | |
hushell | Sorry guys I missed the meeting, the clock didn't rung me up :( | 18:10 |
hushell | lisitsyn: I am reading what you have talked about | 18:10 |
-!- gsomix [~gsomix@109.169.232.211] has joined #shogun | 18:13 | |
gsomix | vodka party? | 18:13 |
gsomix | okay :( | 18:14 |
gsomix | it seems meeting is over | 18:14 |
gsomix | I had problems with electricity in my village | 18:15 |
hushell | gsomix: Don't worry, I didn't in the party as well | 18:21 |
hushell | lisitsyn: about the application/demo, I am going to have an image denoising or image segmentation demo, and work with some real dataset to show the possibility for large scale | 18:23 |
hushell | I'll discuss more with Patrick and send you guys an email | 18:23 |
hushell | Sorry for missing the meeting again! | 18:24 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 18:47 | |
@lisitsyn | hushell: ah don't worry | 18:50 |
hushell | lisitsyn: I went to Seattle for a day and came at 1:00am, this makes me a bit tired out | 18:56 |
hushell | lisitsyn: I will sleep ealier if next meeting is in the morning for me. | 18:57 |
van51 | lisitsyn: sonney is going to remove SGString, right? | 19:19 |
van51 | or if someone else knows | 19:20 |
@lisitsyn | van51: yeah I think so | 19:20 |
van51 | lisitsyn: ok, I'll have my methods use only SGVector then | 19:21 |
van51 | btw, where do you develop your code you guys? | 19:24 |
van51 | right now I'm working on gedit, but I wouldn't mind some auto-complete features | 19:25 |
@lisitsyn | van51: vim + clang_complete | 19:25 |
gsomix | van51, gedit, mc+nano | 19:26 |
votjakovr | van51: emacs + auto-complete-clang | 19:27 |
van51 | thanks for the suggestions! | 19:28 |
van51 | gsomix: do you auto-complete with that configuration? | 19:28 |
gsomix | van51, nope. hardcore mode. | 19:28 |
van51 | gsomix: hehe, that's what I'm trying to avoid :P | 19:28 |
van51 | I think I'll give vim a chance | 19:30 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 19:37 | |
-!- lisitsyn [~lisitsyn@83.234.54.94] has joined #shogun | 20:04 | |
-!- Rasmi [~Rasmi@117.197.244.112] has joined #shogun | 20:51 | |
-!- nube [~rho@49.244.56.141] has quit [Quit: Leaving.] | 20:51 | |
Rasmi | As in the installation manual it is given that after going to the shogun/src directory do ./configure --interfaces=libshogun, that means that libshoguni should be a subfolder of interfaces, but it is not so | 20:54 |
Rasmi | so my query is how to configure libshogun | 20:55 |
Rasmi | there are some static and modular interfaces are given in the shogun/src/interfaces/ folder, then which interface shall i configure to have a c++ interface | 20:57 |
-!- Rasmi [~Rasmi@117.197.244.112] has quit [Ping timeout: 260 seconds] | 21:04 | |
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has left #shogun ["PART #gsoc-gr :QUIT :Leaving."] | 21:25 | |
-!- FSCV [~FSCV@187.210.54.166] has joined #shogun | 21:38 | |
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has left #shogun ["Fallen asleep!"] | 21:41 | |
-!- hushell [~hushell@c-24-21-141-32.hsd1.or.comcast.net] has quit [Ping timeout: 252 seconds] | 21:59 | |
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has joined #shogun | 22:16 | |
--- Log closed Fri Jun 21 00:00:54 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!