--- Log opened Thu Apr 18 00:00:28 2013 | ||
-!- _rohan [~rohan@122.179.42.32] has quit [Quit: Leaving] | 00:18 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 00:48 | |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 01:56 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 02:56 | |
shogun-notifier- | shogun: Viktor Gal :develop * 377fbcb / src/shogun/ui/SGInterface.cpp: https://github.com/shogun-toolbox/shogun/commit/377fbcbb55bcd44e0900489acbd5f79ada0cc982 | 02:56 |
---|---|---|
shogun-notifier- | shogun: Fix SubgradSVM and SubgradLPM removal in SGInterface | 02:56 |
shogun-notifier- | shogun: Viktor Gal :develop * 5bfeb95 / .travis.yml: https://github.com/shogun-toolbox/shogun/commit/5bfeb9560ab6eae751295a7e97e4316a39829a21 | 03:00 |
shogun-notifier- | shogun: Travis: remove commas from package list for ruby | 03:00 |
-!- travis-ci [~travis-ci@ec2-50-16-8-109.compute-1.amazonaws.com] has joined #shogun | 03:24 | |
travis-ci | [travis-ci] it's Viktor Gal'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/6432649 | 03:24 |
-!- travis-ci [~travis-ci@ec2-50-16-8-109.compute-1.amazonaws.com] has left #shogun [] | 03:24 | |
-!- abinash [1261415@gateway/web/cgi-irc/kiwiirc.com/x-xbbpapvsgwwgvloz] has joined #shogun | 03:33 | |
-!- travis-ci [~travis-ci@ec2-50-19-174-217.compute-1.amazonaws.com] has joined #shogun | 03:52 | |
travis-ci | [travis-ci] it's Viktor Gal'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/6432729 | 03:52 |
-!- travis-ci [~travis-ci@ec2-50-19-174-217.compute-1.amazonaws.com] has left #shogun [] | 03:52 | |
-!- abinash [1261415@gateway/web/cgi-irc/kiwiirc.com/x-xbbpapvsgwwgvloz] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 03:53 | |
-!- abinash [1261415@gateway/web/cgi-irc/kiwiirc.com/x-ydlqbmmbqzetsdni] has joined #shogun | 03:55 | |
shogun-notifier- | shogun: Viktor Gal :develop * d53a898 / .travis.yml,src/configure: https://github.com/shogun-toolbox/shogun/commit/d53a8982fa51e129a1e08df11089f8212fca7fa6 | 04:08 |
shogun-notifier- | shogun: Fix ruby call in test_ruby_narray in configure script | 04:08 |
shogun-notifier- | shogun: Try fixing ruby-narray error by enforcing to use ruby1.8 | 04:08 |
-!- abinash [1261415@gateway/web/cgi-irc/kiwiirc.com/x-ydlqbmmbqzetsdni] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 04:11 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 04:21 | |
foulwall | sonney2k: gsomix: I've sent a pull request . Check it pls and I'll start the svm demo | 04:22 |
-!- Yanglittle [deb20afb@gateway/web/freenode/ip.222.178.10.251] has joined #shogun | 04:52 | |
Yanglittle | is there any body? | 05:02 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Ping timeout: 256 seconds] | 05:03 | |
-!- abinash [1261469@gateway/web/cgi-irc/kiwiirc.com/x-muaxsnzmrofdwpxs] has joined #shogun | 05:06 | |
-!- Yanglittle [deb20afb@gateway/web/freenode/ip.222.178.10.251] has quit [Quit: Page closed] | 05:12 | |
-!- abinash [1261469@gateway/web/cgi-irc/kiwiirc.com/x-muaxsnzmrofdwpxs] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 05:30 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Read error: Connection reset by peer] | 05:59 | |
-!- wiking [~wiking@info2k1.hu] has joined #shogun | 05:59 | |
-!- vikram360 [~vikram360@117.192.165.125] has quit [Ping timeout: 245 seconds] | 06:27 | |
-!- sumit_ [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 06:41 | |
-!- lambday [3ba21749@gateway/web/freenode/ip.59.162.23.73] has joined #shogun | 06:53 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 07:08 | |
-!- lambday [3ba21749@gateway/web/freenode/ip.59.162.23.73] has quit [] | 07:12 | |
-!- saket [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun | 07:15 | |
-!- sumit_ [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Ping timeout: 245 seconds] | 07:20 | |
saket | hi | 07:41 |
saket | I am working on making a tree in shogun | 07:41 |
saket | I would like to calculate the dimension in the (CFeatures*) data which has maximum variance | 07:42 |
saket | is there any already function for doing so? | 07:42 |
saket | anyone there? | 07:45 |
-!- vikram360 [~vikram360@117.192.161.30] has joined #shogun | 07:47 | |
-!- k_matsuda [~k_matsuda@h101-111-222-195.catv02.itscom.jp] has quit [Remote host closed the connection] | 07:51 | |
-!- hoijui [~hoijui@dslb-088-074-124-102.pools.arcor-ip.net] has joined #shogun | 08:03 | |
saket | anyone there? | 08:05 |
saket | anyone? | 08:33 |
sonne|work | saket: no there is no function like that | 08:49 |
sonne|work | so this would be a first small welcome patch | 08:50 |
saket | actaully, I needed that function for another patch ;p | 08:50 |
saket | I see that load(CFile) is also not implemented ..Can you please guide me on where (which code) to read about accessing reals from CFeatures | 08:51 |
saket | I see methods like "get_feature_matrix" and "get_feature_vector(index)" as functions in the python interface | 08:57 |
saket | but am i unable to locate them in the src folder | 08:57 |
saket | please help me out with this .. | 08:58 |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:01 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 09:01 | |
saket | sonne|work: I will certainly have get_var() in the patch i submit :-) | 09:04 |
-!- Yanglittle [deb20af9@gateway/web/freenode/ip.222.178.10.249] has joined #shogun | 09:06 | |
sonne|work | saket: it is in CDenseFeatures and SGMatrix | 09:07 |
saket | sonne|work: thanks :) I will have a look | 09:08 |
-!- vikram360 [~vikram360@117.192.161.30] has quit [Ping timeout: 264 seconds] | 09:23 | |
-!- vikram360 [~vikram360@117.192.177.58] has joined #shogun | 09:23 | |
-!- Foulwall [~vkmahfa@221.130.33.37] has joined #shogun | 09:34 | |
Foulwall | Hello guys | 09:35 |
-!- Foulwall [~vkmahfa@221.130.33.37] has quit [Client Quit] | 09:36 | |
@iglesiasg | sonne|work: I guess it is fine if I send the mail of the workshop around in the department where I am doing my master thesis? | 10:11 |
sonne|work | iglesiasg: encouraged even! | 10:13 |
@iglesiasg | cool, I will do it soon then | 10:14 |
sonne|work | wiking: so what do we do about the failing serialization_complex_example.py 1/2 ? | 10:16 |
sonne|work | I have absolutely no idea how to address that | 10:16 |
sonne|work | the only thing we can do is print the objects... | 10:17 |
@iglesiasg | sonne|work: let me know if I can help with this somehow, I don't really know how the serialization works but maybe I could prepare some tests or sth | 10:18 |
-!- saket [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has quit [Ping timeout: 245 seconds] | 10:19 | |
sonne|work | iglesiasg: the problem really is that this test works on all our machines but it does not on the travis buildbot | 10:21 |
sonne|work | absolutely no idea why not | 10:21 |
@iglesiasg | aham I see | 10:21 |
Yanglittle | is there any body? | 10:23 |
@iglesiasg | idd | 10:23 |
Yanglittle | I have a problem , mkl.getsubkernelweights(numkernel), and it always say type error, and what is "int32_t &" is there any body can help me | 10:27 |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has joined #shogun | 10:29 | |
@lisitsyn | Yanglittle: as I said in my mail try it without any parameter | 10:30 |
@lisitsyn | Yanglittle: depends on what mkl here is though | 10:31 |
Yanglittle | i tried , it expected a argument. | 10:33 |
@iglesiasg | Yanglittle: so you are using python_modular, don't you? | 10:36 |
Yanglittle | yes. | 10:36 |
@iglesiasg | Yanglittle: I think it can be a problem with the type, that you have to say explicitily that what you are giving is an int32_t | 10:37 |
@iglesiasg | but it is hard to tell if I cannot see the code you are using and the exact error | 10:37 |
Yanglittle | yeah ,ok I run the mkl_multiclass_modualr and add the code "mkl.getsubkernelweights(3)" after mkl.train() | 10:40 |
Yanglittle | it says: The debugged program raised the exception TypeError "in method 'MKLMulticlass_getsubkernelweights', argument 2 of type 'int32_t &'" File: /home/liyang/shogun/examples/documented/python_modular/mkl_multiclass_modular.py, Line: 75 | 10:40 |
@iglesiasg | Yanglittle: ok let me try it here | 10:40 |
@lisitsyn | ahh | 10:42 |
@lisitsyn | Yanglittle: try mkl.get_kernel().get_subkernel_weights() | 10:43 |
@lisitsyn | Yanglittle: or rather | 10:44 |
@lisitsyn | just call get_subkernel_weights() on your combined kernel | 10:44 |
@iglesiasg | lisitsyn: it is because of the & thing, no way to do that in python I think | 10:44 |
@lisitsyn | iglesiasg: it is legacy from the days we had no sgvector | 10:45 |
@iglesiasg | lisitsyn: mmm why to use & when there is no sgvector? | 10:45 |
@iglesiasg | it is not like passing a pointer, int32_t* | 10:45 |
@iglesiasg | probably I am getting something wrong | 10:45 |
@lisitsyn | iglesiasg: yes there should be a pointer too but the spirit is the same :) | 10:46 |
@iglesiasg | lisitsyn: mmm I don't understand | 10:46 |
@lisitsyn | iglesiasg: before sgvector getters were like | 10:47 |
@lisitsyn | (float64_t**, int32_t&) | 10:47 |
@iglesiasg | I understood the int32_t numweights as an int by reference | 10:47 |
Yanglittle | ok, now it can works, thanks very much! | 10:47 |
@iglesiasg | ah ok I see it now | 10:47 |
@iglesiasg | the int32_t& is to output the size of the float64_t* returned | 10:47 |
@lisitsyn | yes | 10:49 |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 10:51 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has joined #shogun | 11:00 | |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 11:11 | |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 11:12 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 11:12 | |
-!- sumit_ [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 11:14 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 11:15 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has joined #shogun | 11:17 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has quit [Client Quit] | 11:21 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has joined #shogun | 11:24 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has quit [Client Quit] | 11:25 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has joined #shogun | 11:26 | |
-!- abinash [75ef5e6e@gateway/web/cgi-irc/kiwiirc.com/ip.117.239.94.110] has quit [Client Quit] | 11:28 | |
-!- shogun-toolbox [~shogun@7nn.de] has quit [Ping timeout: 256 seconds] | 11:31 | |
--- Log closed Thu Apr 18 11:31:44 2013 | ||
--- Log opened Thu Apr 18 11:33:03 2013 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 11:33 | |
-!- Irssi: #shogun: Total of 19 nicks [3 ops, 0 halfops, 0 voices, 16 normal] | 11:33 | |
-!- Irssi: Join to #shogun was synced in 66 secs | 11:33 | |
-!- vikram360 [~vikram360@117.192.177.58] has quit [Ping timeout: 252 seconds] | 11:36 | |
-!- HeikoS [~heiko@nat-187-85.internal.eduroam.ucl.ac.uk] has joined #shogun | 11:39 | |
sonne|work | hei HeikoS - please send around all the CfPs to whoever you can think of :D | 11:42 |
-!- shogun-toolbox [~shogun@7nn.de] has quit [Ping timeout: 245 seconds] | 11:42 | |
--- Log closed Thu Apr 18 11:42:24 2013 | ||
--- Log opened Thu Apr 18 11:42:30 2013 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 11:42 | |
-!- Irssi: #shogun: Total of 19 nicks [3 ops, 0 halfops, 0 voices, 16 normal] | 11:42 | |
-!- Irssi: Join to #shogun was synced in 6 secs | 11:42 | |
HeikoS | sonne|work: will do :) | 11:44 |
-!- hoijui [~hoijui@dslb-088-074-124-102.pools.arcor-ip.net] has quit [Ping timeout: 246 seconds] | 11:45 | |
-!- k_matsuda [~k_matsuda@h101-111-222-195.catv02.itscom.jp] has joined #shogun | 11:51 | |
@lisitsyn | HeikoS: is there any difference between berlin airports? | 12:21 |
HeikoS | lisitsyn: better ask sonney2k | 12:21 |
@lisitsyn | sonne|work: ^ | 12:21 |
HeikoS | Dont know | 12:21 |
HeikoS | always go there by car :) | 12:21 |
@lisitsyn | HeikoS: where are you going to stay btw? | 12:22 |
HeikoS | lisitsyn: dont know, probably friends | 12:22 |
HeikoS | or a hostel | 12:22 |
HeikoS | I can let you know places, but again, sonney2k probably knows the place a bit better | 12:23 |
@lisitsyn | HeikoS: any estimate on cost of staying there? | 12:23 |
HeikoS | lisitsyn: I never stayed in a hotel there | 12:23 |
HeikoS | oh yes once | 12:23 |
@lisitsyn | heh | 12:23 |
HeikoS | a very nice hostel | 12:23 |
HeikoS | but I forgot the price ;) | 12:23 |
HeikoS | not too much | 12:23 |
@lisitsyn | yay germany visa is >200$ for me | 12:23 |
@lisitsyn | :D | 12:23 |
HeikoS | wow | 12:23 |
@lisitsyn | HeikoS: I should also prove I have enough money to get back :D | 12:28 |
HeikoS | lisitsyn: haha .) | 12:29 |
HeikoS | do you? :D | 12:30 |
@lisitsyn | HeikoS: I hope so! | 12:30 |
wiking | sonne|work: yo | 12:31 |
wiking | sonne|work: i have no idea why that python example fails | 12:31 |
wiking | lisitsyn: you need sponsors? :D | 12:33 |
@lisitsyn | wiking: like sponsor that would take care of me? | 12:34 |
@lisitsyn | for visa? | 12:34 |
wiking | yes | 12:35 |
wiking | for visa | 12:35 |
wiking | or you are going with an agency? | 12:35 |
@lisitsyn | wiking: I am undecided - will take care of this in next few days | 12:35 |
wiking | lisitsyn: if u need an invitation letter let me know | 12:36 |
@lisitsyn | wiking: but you can't invite me to germany right? :) | 12:36 |
wiking | i can | 12:37 |
wiking | i mean not to germany | 12:37 |
wiking | but to EU | 12:37 |
wiking | i mean schengen | 12:37 |
wiking | which is pretty much the same | 12:37 |
-!- sumit_ [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Ping timeout: 245 seconds] | 12:37 | |
@lisitsyn | wiking: I just checked - I have to have some place to stay to get visa | 12:38 |
wiking | i mean if u get the visa (schengen) you would be able to travel within schengen zone | 12:38 |
@lisitsyn | yes I know | 12:38 |
wiking | yeah | 12:38 |
wiking | well i dont know if sonne|work can offer you a place | 12:38 |
wiking | but if that's too complicated | 12:38 |
@lisitsyn | hah I guess no | 12:39 |
wiking | we can tell that you are going to stay at my place in budapest | 12:39 |
wiking | and that's all | 12:39 |
wiking | you get the visa for hungary, which is basically a visa for germany as well | 12:39 |
wiking | and then it should be all good | 12:40 |
@lisitsyn | I see | 12:40 |
wiking | if u want one of these days i can call the russian embassy in budapest | 12:40 |
wiking | to ask them for most of the things... | 12:40 |
@lisitsyn | you are helpful! thanks :) | 12:40 |
wiking | but maybe - if u decide to go this way - you should call the hungarian embassy in russia | 12:40 |
wiking | just to be sure about the papers we need to provide them | 12:41 |
@lisitsyn | I'll reach some agency and ask them | 12:41 |
wiking | lisitsyn: +7-495-796-9370 | 12:41 |
@lisitsyn | wiking: where are you going to stay in berlin btw? | 12:41 |
HeikoS | lisitsyn, wiking isn't it easier to do all this in Germany? I mean get a place to stay there? | 12:41 |
HeikoS | lisitsyn: Once you booked a hostel you have a place to stay | 12:41 |
wiking | HeikoS: hehe it's not that easy | 12:42 |
wiking | ;) | 12:42 |
HeikoS | sonney2k can give you a list | 12:42 |
wiking | HeikoS: it's much better if he can tell to the embassy | 12:42 |
HeikoS | ok | 12:42 |
wiking | that he is staying at some persons private place | 12:42 |
HeikoS | well then, go ahead :) | 12:42 |
HeikoS | why? | 12:42 |
wiking | well | 12:42 |
wiking | because EU is an asshole | 12:42 |
wiking | ..? :) | 12:42 |
@lisitsyn | HeikoS: yes - but I just don't know if they provide proper papers for german visa | 12:42 |
wiking | and they are even more assholes with russians...? :) | 12:42 |
@lisitsyn | wiking: haha | 12:42 |
HeikoS | annoying | 12:42 |
wiking | sorry but this is the truth | 12:43 |
wiking | once we had to apply for a schengen zone visa for a friend of mine | 12:43 |
wiking | who was from middle-east | 12:43 |
wiking | it was hell | 12:43 |
wiking | i was waiting when are they going to ask for a dna sample of my friend | 12:43 |
@lisitsyn | hahah | 12:43 |
wiking | and on the end by 'their mistake' they refused to give him the visa | 12:43 |
wiking | ... | 12:43 |
wiking | lisitsyn: if the hungarian embassy in russia | 12:44 |
wiking | can provide schengen visa | 12:44 |
wiking | then you are afaik much better off than germans | 12:44 |
wiking | just because of the history between the 2 countries | 12:44 |
@lisitsyn | wiking: heh | 12:44 |
@lisitsyn | yeah may be | 12:44 |
wiking | you know old communist comrades ;) | 12:44 |
wiking | but then again | 12:45 |
wiking | in some countries they cannot issue schengen visas | 12:45 |
wiking | and it's done by another embassy | 12:45 |
wiking | lisitsyn: so if u can please call these guys in moscow | 12:45 |
wiking | and ask them | 12:45 |
@lisitsyn | wiking: HeikoS: I am asking where you are going to stay as we may book the same hotel? | 12:45 |
wiking | lisitsyn: i have 0 ideas yet | 12:46 |
wiking | i think i'm going to fix this in may some time | 12:46 |
@lisitsyn | wiking: my plan is to check about direct -> germany thing and then compare | 12:46 |
@lisitsyn | wiking: so you have hungarian citizenship rite? | 12:47 |
wiking | lisitsyn: as well, yes | 12:47 |
@lisitsyn | wiking: you have some more? :) | 12:48 |
wiking | lisitsyn: yes seribna | 12:48 |
@lisitsyn | ahh | 12:48 |
wiking | *serbian | 12:48 |
wiking | so yeah for the hungarians we can provide all stuff like: i am a 'sponsor' of your stay during your visit and that you can stay at my flat during the time your visit | 12:49 |
wiking | anyhow somebody has any ideas with the serialization issue on travis? | 12:50 |
@lisitsyn | that's cool | 12:50 |
wiking | i saw it once failed on buildbot as well | 12:50 |
@lisitsyn | wiking: thanks! let us consider that if germany deny me :D | 12:50 |
wiking | lisitsyn: okok just be on time | 12:50 |
wiking | lisitsyn: it does take them some fair amount of time to decide anything | 12:51 |
@lisitsyn | yeah I should take care of that next 2 weeks or I could get out of time | 12:51 |
wiking | lisitsyn: yeah i really advise you to start it latest next week | 12:51 |
@lisitsyn | I wanted to start early but postponed postponed postponed | 12:51 |
wiking | for my friend it took them ages to do anything | 12:51 |
wiking | and then on the end they denied | 12:51 |
wiking | and 1.5 week before his trip | 12:51 |
@lisitsyn | wiking: some agencies here say they can make a german visa for 7500 roubles | 12:52 |
wiking | they told him that 'aaaahh we made a mistake'.... | 12:52 |
wiking | but it was too late | 12:52 |
wiking | and he couldnt do the trip anymore | 12:52 |
wiking | lisitsyn: how much is that in usd? | 12:52 |
@lisitsyn | divide by 30 | 12:52 |
wiking | ohh shiiiiiit | 12:52 |
wiking | that's craaazy expensive | 12:52 |
@lisitsyn | wiking: but I wouldn't have to go to moscos | 12:52 |
@lisitsyn | moscow | 12:52 |
wiking | aaah otherwise you *have* to? | 12:52 |
@lisitsyn | wiking: yeah may be | 12:53 |
wiking | sure? | 12:53 |
wiking | i mean i know this friend of mine did it via mail | 12:53 |
@lisitsyn | wiking: german embassy says they want an interview | 12:53 |
wiking | btw: any of these places are nearer to you than moscow: http://hungary.visahq.com/embassy/Russia/ | 12:53 |
@lisitsyn | wiking: I am just like a cluster in k-means | 12:54 |
@lisitsyn | all of these are ~1000 km :D | 12:54 |
wiking | shitfuckers | 12:54 |
wiking | ah yeah you are in samara | 12:54 |
wiking | ok | 12:54 |
wiking | call that number i gave you | 12:54 |
@lisitsyn | wiking: yes, toliatty currently | 12:54 |
wiking | just to ask them | 12:54 |
wiking | what are their requirements | 12:55 |
wiking | maybe they dont want an interview with you | 12:55 |
@lisitsyn | wiking: alright | 12:55 |
wiking | lisitsyn: as far as i understood from some papers | 12:57 |
wiking | you as a russian citizen | 12:57 |
wiking | have some 'light' criterions for getting the schengen visa | 12:58 |
@lisitsyn | yeah it must be much easier than for your friend from middle east | 12:58 |
wiking | heheheh | 12:59 |
wiking | HeikoS: ping | 13:01 |
HeikoS | wiking: | 13:02 |
wiking | HeikoS: wtf is happening in that serialization script? | 13:02 |
HeikoS | wiking: which one? | 13:02 |
wiking | complex serialization...py | 13:02 |
HeikoS | wiking: dont know, did I write that? | 13:03 |
wiking | dunno | 13:03 |
HeikoS | this one keeps on failing on my local machine | 13:04 |
HeikoS | sonney2k recently did some fixes for serialization | 13:04 |
HeikoS | haven't looked into it for a while | 13:04 |
wiking | btw: anybody knows what's the hard limit for a repository size on github | 13:06 |
wiking | ? | 13:06 |
@lisitsyn | wiking: I have seen some info where they say 1gb is soft limit | 13:07 |
@lisitsyn | and when you reach 1gb they ask you to reduce the size | 13:07 |
wiking | aho | 13:07 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 13:11 | |
wiking | that should be enough for me.. thnx lisitsyn | 13:13 |
foulwall | sonney2k: Hey. maybe some more concrete suggests on my new pull request? I'm refactoring the code. | 13:15 |
@lisitsyn | wiking: okay other agency says 80EUR for visa | 13:22 |
@lisitsyn | :D | 13:22 |
@lisitsyn | https://maps.google.com/maps?q=+52.376552,+5.198303&hl=en&ll=52.37655,5.198335&spn=0.000119,0.000312&sll=35.101934,-95.712891&sspn=41.358418,81.914062&t=h&z=22 | 13:30 |
-!- Yanglittle [deb20af9@gateway/web/freenode/ip.222.178.10.249] has quit [Ping timeout: 245 seconds] | 13:34 | |
wiking | lisitsyn: heheh yeah i saw that yesterday or something | 13:41 |
wiking | btw we should have unit testing as well on travis i guess | 13:43 |
HeikoS | wiking: yes, that would be very cool! | 13:46 |
wiking | HeikoS: it's not complicated | 13:48 |
wiking | the only problem is that somehow it shoudl be done | 13:48 |
HeikoS | yeah as always :) | 13:48 |
wiking | only once | 13:48 |
HeikoS | I like this travis stuff | 13:49 |
wiking | so only for the libshogun run | 13:49 |
wiking | not for all the modular cases | 13:49 |
HeikoS | yes of course | 13:49 |
wiking | which makes it of course a bit more complex :P | 13:49 |
HeikoS | we have the modular examples, that should be fine | 13:49 |
HeikoS | unit tests with valgrind would also be nice :) | 13:49 |
wiking | HeikoS: well that's only one more package :P | 13:50 |
HeikoS | yeah :) | 13:50 |
wiking | HeikoS: what's the command for running that? :))) | 13:51 |
wiking | HeikoS: i know i've written it but still | 13:51 |
HeikoS | wiking: | 13:51 |
HeikoS | no idea | 13:51 |
HeikoS | I usually do cat Makefile | grep valgrind :D | 13:51 |
wiking | lol | 13:51 |
HeikoS | I try hard *not* to remember those things in order to keep my brain tidy ;) | 13:52 |
sonne|work | foulwall: what do you mean? | 13:52 |
wiking | anyhow first i debug something because i dont get some shit... | 13:52 |
HeikoS | okay | 13:53 |
foulwall | sonne|work: I am trying to make the code as most as reusable as I can. | 13:54 |
sonne|work | foulwall: I guess the three parts: getting input / shogun learning / applying & output should be enough? | 13:55 |
foulwall | I wrote a base class called train&learn model | 13:56 |
foulwall | clustering and svr is derived from the train&learn | 13:57 |
foulwall | but the split up is not easy to do. I'm trying to find a general method. | 13:58 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Ping timeout: 252 seconds] | 14:08 | |
-!- foulwall [~foulwall@li379-21.members.linode.com] has joined #shogun | 14:08 | |
foulwall | sonne|work: Can you see me? | 14:09 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 14:20 | |
wiking | lol | 14:21 |
wiking | i cannot unref an svm | 14:21 |
@lisitsyn | wiking: why? | 14:23 |
@lisitsyn | it frightens you? | 14:23 |
-!- gzhd79 [~zeller@embln.embl.de] has joined #shogun | 14:27 | |
sonne|work | foulwall: what do you mean? | 14:31 |
sonne|work | lisitsyn: btw do you send out confirmation emails or anything to the people registering for the workshop? | 14:33 |
@lisitsyn | sonne|work: will send a bunch today | 14:33 |
@lisitsyn | sonne|work: who are subscribers of that address? | 14:34 |
@lisitsyn | internal? should I CC it? | 14:34 |
sonne|work | you and me so far | 14:34 |
foulwall | sonne|work: do we need a model for all demo which contains an train and apply process or just split the function | 14:37 |
sonne|work | foulwall: just split it for now. However data input is very similar | 14:38 |
foulwall | sonne|work: Ok. | 14:38 |
-!- vikram360 [~vikram360@117.192.163.136] has joined #shogun | 14:49 | |
wiking | lisitsyn: for binary svm which has better performance (speed): libsvm or svmlight, or it all depends? | 14:51 |
@lisitsyn | wiking: kernel? | 14:51 |
wiking | yep | 14:51 |
wiking | btw i might need to test as well liblinear + HKM :P | 14:52 |
@lisitsyn | it depends of course but I'd expect svmlight could be faster | 14:52 |
-!- deerishi [73f88294@gateway/web/freenode/ip.115.248.130.148] has joined #shogun | 14:53 | |
@lisitsyn | wiking: HKM + liblinear rocks :) | 14:53 |
@lisitsyn | wiking: at least it makes you able to find better features / settings | 14:53 |
gsomix | hi | 14:58 |
-!- foulwall [~foulwall@li379-21.members.linode.com] has quit [Remote host closed the connection] | 15:30 | |
-!- deerishi [73f88294@gateway/web/freenode/ip.115.248.130.148] has quit [Ping timeout: 245 seconds] | 15:32 | |
-!- deerishi [73f88294@gateway/web/freenode/ip.115.248.130.148] has joined #shogun | 15:39 | |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 15:40 | |
sonne|work | wiking: it really depends... in my experience if you have lots of data from one class and not so much from the other svmlight is faster | 15:47 |
wiking | sonne|work: it's kind of like equaly distributed | 15:48 |
sonne|work | svm light better benefits multiple cores | 15:50 |
sonne|work | but otherwise libsvm might be a bit faster | 15:50 |
sonne|work | maybe more stable too | 15:50 |
deerishi | HeikoS: u told me to implement a scik it example on KPCA,could you guide me a bit on how to go about it.i have been trying the shogun KPCA library,but so far did not get anything good | 15:52 |
HeikoS | deerishi: hi! | 15:52 |
HeikoS | deerishi: what I meant is just to use the data in the scikit learn and produce similar looking plots | 15:52 |
HeikoS | using shogun's KPCA implementation | 15:52 |
HeikoS | your code is fine | 15:52 |
HeikoS | but I would use a different example | 15:52 |
HeikoS | the scikit example is very nice since it illustrates the method very nicely | 15:53 |
deerishi | just one quick question : shogun's does explicitly gives back the modified data KPCA ?the code i wrote gave back the densities of the data on a line | 15:54 |
deerishi | ohh.i think i have an idea,will try it out and let you know | 15:56 |
deerishi | HeikoS:thanks | 15:56 |
HeikoS | deerishi: no | 15:56 |
HeikoS | it gives you a matrix | 15:56 |
HeikoS | which is DxN where D are the number of eigevectors you use | 15:56 |
HeikoS | you just have to plot carefully | 15:56 |
HeikoS | pairs of points | 15:57 |
HeikoS | deerishi: you should be almost there! :) | 15:57 |
deerishi | i hv an idea,will work on it tonight.thanks!! | 15:58 |
HeikoS | deerishi: thanks for keeping on it, see you! | 15:58 |
-!- gsomix_ [~gsomix@37.61.181.37] has joined #shogun | 15:58 | |
-!- thoralf [~thoralf@enki.zib.de] has joined #shogun | 16:00 | |
thoralf | Hey. | 16:00 |
sonne|work | Ho :) | 16:01 |
-!- gsomix [~gsomix@37.61.179.101] has quit [Ping timeout: 245 seconds] | 16:01 | |
-!- gzhd79 [~zeller@embln.embl.de] has quit [Quit: gzhd79] | 16:03 | |
-!- gzhd79 [~zeller@embln.embl.de] has joined #shogun | 16:06 | |
wiking | lisitsyn: svmsgd seems to be cool ;) | 16:07 |
wiking | i mean much faster for my dataset :P | 16:07 |
sonne|work | wiking: well it just does x passes over the data | 16:11 |
sonne|work | so no wonder | 16:11 |
wiking | mmmm but unfortunately all the options | 16:12 |
wiking | are way worse than random forest :( | 16:12 |
wiking | i guess we dont have random forest, yet :P | 16:12 |
wiking | woah | 16:14 |
wiking | pnorm preprocessor segfaults | 16:14 |
sonne|work | thoralf: better chat in the channel :) | 16:15 |
sonne|work | yeah I know I did write the hashing stuff in shogun | 16:15 |
sonne|work | wiking: backtrace? | 16:15 |
sonne|work | hard to believe since it is so simple | 16:16 |
wiking | doh | 16:23 |
wiking | something is really weird | 16:24 |
wiking | i have an SGMatrix | 16:24 |
wiking | and with memcpy i fill it up | 16:24 |
wiking | and right away after memcpy i can even print out the right values | 16:24 |
wiking | but then when i do on the whole matrix | 16:24 |
wiking | a display_matrix() | 16:24 |
wiking | it's a full 0 matrix | 16:24 |
wiking | i mean the output | 16:24 |
wiking | any ideas? | 16:26 |
-!- sids_aquarius [~sidi_@li400-124.members.linode.com] has joined #shogun | 16:30 | |
wiking | ok seem display_matrix is aint working | 16:30 |
sonne|work | huh? | 16:33 |
sonne|work | I used that and it worked | 16:33 |
wiking | coz it perfectly works if i put that matrix into a densefeatures class and then do a 1-by-1 feats->get_feature_vector(i); display_vector() | 16:33 |
wiking | but if i just do a display_matrix it's like as if the matrix is a big 0 | 16:34 |
sonne|work | wiking: please put some test code somewhere in a github issue and mark it entrance... | 16:35 |
wiking | sonne|work: ok will do | 16:35 |
vgorbati | hi all, I am a student wisihing to participate in GSoC 2013. To get familiar with shogun I have forked, cloned the shogun-git workspace and compiled it. To continue, I want to try working on the issue https://github.com/shogun-toolbox/shogun/issues/968. Can anyone talk with me about it? | 16:37 |
wiking | now more the question is | 16:37 |
wiking | that why pnorm dies | 16:37 |
wiking | :) | 16:37 |
wiking | sonne|work: it works fine with a small SGMatrix | 16:38 |
wiking | and it dies with a big one | 16:38 |
sonne|work | big == ? | 16:38 |
wiking | 380000x18 | 16:38 |
wiking | on a 19*18 | 16:38 |
wiking | it works fine | 16:38 |
thoralf | sonne: Yes, I was wondering... since your name appears in the copyright notice of all the hashing kernels. ;) | 16:40 |
wiking | sonne|work: hahaha i think i know why, although not 100% | 16:43 |
wiking | sonne|work: u reckon this could happen when i apply PNorm (which works with fixed float64_t) on a DenseFeatures<float32_t> ? | 16:43 |
wiking | i mean memory access problems | 16:43 |
wiking | imo PNorm should be templated as well | 16:44 |
-!- gsomix_ [~gsomix@37.61.181.37] has quit [Ping timeout: 272 seconds] | 16:55 | |
thoralf | I am trying to process text data with shogun, but the examples I found only deal with DNA data, producing one feature per CHAR. | 16:57 |
thoralf | Can anyone give a pointer to a piece of code with WORD features? :) | 16:58 |
thoralf | The class StringWordFeatures pretends to be the right one, but I did not find out how to use it. | 16:59 |
thoralf | (The class name within the python bindings.) | 17:01 |
-!- deerishi [73f88294@gateway/web/freenode/ip.115.248.130.148] has quit [Ping timeout: 245 seconds] | 17:04 | |
sonne|work | thoralf: yeah that would be task too - we don't have features for text | 17:12 |
sonne|work | the word in there just denotes 2 bytes | 17:12 |
thoralf | sonne: Thanks - at least I don't need to search any more. | 17:14 |
thoralf | sonne: A preprocessing using NLTK wouldn't be hard, so I think this might be the way to go. | 17:15 |
wiking | sonne|work: when i template a class why i dont have m_parameter ? :) | 17:19 |
sonne|work | wiking: what? | 17:26 |
-!- gzhd79 [~zeller@embln.embl.de] has quit [Quit: gzhd79] | 17:27 | |
sonne|work | thoralf: it would be nice to have some 'text document' -> hashed list of words / n-grams in shogun though | 17:27 |
wiking | sonne|work: https://gist.github.com/vigsterkr/5413663 | 17:30 |
wiking | sonne|work: any comments on this diff? | 17:30 |
thoralf | sonne: Parsing text in C/C++ is not the funniest thing to do, so actually I was trying to avoid this part. ;) | 17:33 |
-!- sumit [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 17:33 | |
-!- gzhd79 [~zeller@embln.embl.de] has joined #shogun | 17:35 | |
-!- foulwall [~foulwall@2001:da8:215:c252:193a:a95f:9fad:517a] has joined #shogun | 17:40 | |
foulwall | Hello dudes. | 17:40 |
wiking | sonne|work: the only thing is that in this case i cannot get SG_ADD working | 17:40 |
sonne|work | foulwall: hello | 17:41 |
sonne|work | thoralf: yes indeed but I guess there is no way around this | 17:41 |
foulwall | sonney2k: I've pulled a request. check it pls. I'm wondering the reliability of the code. | 17:42 |
sonne|work | wiking: why do you need the templated thing? | 17:42 |
wiking | sonne|work: well this is the thing | 17:42 |
wiking | sonne|work: i've got segfault | 17:42 |
wiking | sonne|work: when i applied the current PNorm on a DenseFeatures<float32_t> | 17:42 |
wiking | sonne|work: since of course sizeof(float32_t) < sizeof(float64_t) | 17:43 |
wiking | sonne|work: of course the best would be that the user doesn't have to give | 17:43 |
wiking | the PNorm<T> | 17:43 |
wiking | as it should be known where it's applied | 17:43 |
wiking | i mean the T type in the Features | 17:44 |
wiking | right? | 17:44 |
wiking | so this should be somehow detected runtime :P | 17:44 |
sonne|work | foulwall: looks much more readable now | 17:44 |
@lisitsyn | why do we need templated features/preprocessors/etc exposed to the interfaces? | 17:44 |
wiking | lisitsyn: any idea how to solve this in case of PNorm? | 17:44 |
foulwall | sonne|work: I'm looking for a way to hide the related but not public method in the class. | 17:44 |
@lisitsyn | wiking: no - I am ok with that change but we are going bad | 17:45 |
sonne|work | wiking: it is quite weird already that you were able to attach this processor to your float32_t data | 17:45 |
sonne|work | there should be a check in there | 17:45 |
wiking | lisitsyn: indeed | 17:45 |
thoralf | sonne: Mapping all text features (unigrams, n-grams) to numbers and storing them in sparse feature files would be a good start. | 17:45 |
@lisitsyn | wiking: have you seen my crazy ideas on API? | 17:45 |
sonne|work | foulwall: my suggestion would be to split up the file - I mean the svr and clustering class could be in some other class or so | 17:46 |
foulwall | Oh, I see. | 17:46 |
thoralf | sonne: Since you probably want to implement all this text-processing-stuff (like stopword removal, stemming) in shogun, this could easily be outsourced to existing tools. | 17:46 |
foulwall | I'm fixing that. | 17:46 |
sonne|work | thoralf: but then what would you need shogun for? | 17:47 |
sonne|work | I mean computing a hash is what you could outsource too | 17:47 |
thoralf | sonne: MKL, Hashing? | 17:47 |
thoralf | sonne: Yes and no: When outsourcing the hash, you don't hav influence to the hashing function any more. | 17:48 |
thoralf | while learning. | 17:48 |
sonne|work | computing hash on n-grams is easy to do in C++ at least | 17:48 |
thoralf | Ack. | 17:48 |
sonne|work | computing hashes on words (with some stop tokens) too | 17:48 |
sonne|work | stopword removal / stemming is a different story | 17:49 |
sonne|work | foulwall: remove the old pull request then and create just one pull request with the split up code | 17:50 |
foulwall | sonne|work: ok. | 17:50 |
sonne|work | wiking: look at shogun/features/DenseFeatures.cpp - it uses SG_ADD | 17:50 |
sonne|work | lisitsyn: we indeed mostly only have preprocessors for double/float | 17:50 |
sonne|work | wiking: I would suggest to have this p-norm preproc just for double & float | 17:51 |
@lisitsyn | sonne|work: I tend to think we should get rid of templates here | 17:53 |
@lisitsyn | it is not possible to get rid of them at all but they should go deeper | 17:53 |
@lisitsyn | like in policy-based design (scnr) | 17:54 |
-!- gzhd79 [~zeller@embln.embl.de] has quit [Quit: gzhd79] | 18:05 | |
-!- foulwall [~foulwall@2001:da8:215:c252:193a:a95f:9fad:517a] has quit [Ping timeout: 245 seconds] | 18:08 | |
wiking | sonne|work: for some reason i get this | 18:15 |
wiking | preprocessor/PNorm.cpp:114:2: error: use of undeclared identifier 'm_parameters' | 18:16 |
wiking | SG_ADD(&m_p, "norm", "P-norm parameter", MS_NOT_AVAILABLE); | 18:16 |
wiking | ^ | 18:16 |
wiking | ../shogun/base/SGObject.h:65:3: note: expanded from macro 'SG_ADD' | 18:16 |
wiking | m_parameters->add(param, name, description);\ | 18:16 |
wiking | ^ | 18:16 |
wiking | preprocessor/PNorm.cpp:114:2: error: use of undeclared identifier 'm_model_selection_parameters' | 18:16 |
wiking | SG_ADD(&m_p, "norm", "P-norm parameter", MS_NOT_AVAILABLE); | 18:16 |
wiking | ^ | 18:16 |
wiking | ../shogun/base/SGObject.h:67:4: note: expanded from macro 'SG_ADD' | 18:16 |
wiking | m_model_selection_parameters->add(param, name, description);\ | 18:16 |
wiking | using SG_ADD | 18:16 |
wiking | sonne|work: if i don't define template class CPNorm<bool>; etc. | 18:17 |
wiking | sonne|work: linker starts to cry that it cannot find | 18:17 |
wiking | the reference on those | 18:17 |
wiking | and for some reaseon they are referenced from class_list.cpp.o | 18:18 |
wiking | any magic tricks to to avoid this error? | 18:18 |
sonne|work | wiking: yeah you are deriving from a templated class | 18:18 |
sonne|work | it is a massacre | 18:19 |
sonne|work | so just create another PNorm preproc for floats | 18:19 |
sonne|work | and relax | 18:19 |
wiking | sonne|work: hahhahaha | 18:19 |
wiking | sonne|work: or have a PNorm preproc now for all? | 18:20 |
sonne|work | otherwise IIRC you have to specify which type of the base class you are referring to | 18:20 |
wiking | because that's pretty fast | 18:20 |
wiking | i mean implementation wise | 18:20 |
wiking | as it's already ready on my machine | 18:20 |
wiking | anyways i have this | 18:21 |
wiking | in init | 18:21 |
wiking | ASSERT(features->get_feature_class()==C_DENSE) ASSERT(features->get_feature_type()==F_DREAL) | 18:21 |
wiking | so it won't run on any other type of features | 18:21 |
wiking | just floats | 18:21 |
sonne|work | ok | 18:21 |
sonne|work | double | 18:21 |
sonne|work | ! | 18:21 |
wiking | float and double yeah | 18:21 |
wiking | the only question now is that why's that error for SG_ADD | 18:22 |
wiking | ok it's funny if i specify like this the function: template <class ST> void CPNorm<ST>::register_param () | 18:25 |
wiking | i mean template <> void CPNorm<float32_t>::register_param () | 18:25 |
wiking | then it works but of course it'll cry that for the other possible template paramaters this function (register_param) does not exist | 18:25 |
wiking | but at least it doesn't cry for not having m_parameters and m_model_selection_parameters variables | 18:26 |
wiking | i have to leave if you guys have a sane soluation for this problem please write it here i'll read the logs and then commit this fix | 18:28 |
sonne|work | wiking: yes as I said it needs the actual type | 18:33 |
wiking | sonne|work: so i should put there by hand all the type? :) | 18:33 |
wiking | ok i see | 18:33 |
sonne|work | so it will work in the specialization | 18:33 |
wiking | will do when i'm online again | 18:33 |
wiking | bbl | 18:33 |
sonne|work | that is why I dont' like deriving templated classes from templated classes | 18:33 |
-!- gsomix [~Miranda@46.20.65.215] has joined #shogun | 18:34 | |
gsomix | hi | 18:34 |
-!- gzhd79 [~zeller@mnhm-4d0128b3.pool.mediaWays.net] has joined #shogun | 18:35 | |
sonne|work | ho | 18:39 |
-!- sumit [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Ping timeout: 245 seconds] | 18:39 | |
gsomix | sonne|work: :) how are you? | 18:40 |
sonne|work | *overload* | 18:40 |
gsomix | sonne|work: hang in there | 18:43 |
-!- gzhd79 [~zeller@mnhm-4d0128b3.pool.mediaWays.net] has quit [Quit: gzhd79] | 18:43 | |
@lisitsyn | sonne|work: try to not hang yourself! | 18:49 |
-!- utsavjain [~tiripS@14.139.82.6] has joined #shogun | 18:53 | |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has joined #shogun | 18:54 | |
dvalcarce | hi everyone! | 18:54 |
utsavjain | Hi | 18:55 |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has quit [Remote host closed the connection] | 18:56 | |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has joined #shogun | 18:57 | |
utsavjain | Hi, I like to participate in GSOC'13 with your organization. I am interested in Machine learning and its tool. What is the first steps, I should follow | 18:57 |
utsavjain | ? | 18:57 |
dvalcarce | sorry for not connecting before, i have taken a lot of subjects this semester | 18:58 |
dvalcarce | i am working on integrating my demos with django right now | 18:58 |
@lisitsyn | utsavjain: subscribe to the mailing list and check archives of the mailing list - you may find some information there | 18:59 |
utsavjain | Thanks! :) | 19:00 |
sonne|work | dvalcarce: you are david? | 19:01 |
dvalcarce | i am daniel valcarce | 19:01 |
dvalcarce | i think i sent you an email two days ago | 19:01 |
sonne|work | daniel yes | 19:02 |
sonne|work | dvalcarce: you did this cool binary svm with gaussian kernel heatmap demo right? | 19:02 |
dvalcarce | haha, yes :) | 19:02 |
sonne|work | dvalcarce: so just hang in here if you can | 19:03 |
dvalcarce | sonne|work: ok, no problem | 19:04 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 245 seconds] | 19:08 | |
dvalcarce | sonne|work: there is a regression demo at <server>/demo/svr/entrance right now, should i add my demos at <server>/demo/binary/entrance and <server>/demo/multi/entrance urls? | 19:19 |
sonne|work | dvalcarce: demo/classification/{binary,multiclass} ? | 19:21 |
sonne|work | sounds better IMHO | 19:22 |
dvalcarce | ok | 19:23 |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 19:27 | |
-!- HeikoS [~heiko@nat-187-85.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 19:29 | |
-!- hoijui [~hoijui@dslb-088-074-124-102.pools.arcor-ip.net] has joined #shogun | 19:30 | |
@lisitsyn | sonne|work: just a quick thing - welcome letter should contain (we would be happy to see you, ask us if have questions) | 19:36 |
@lisitsyn | sonne|work: anything else? | 19:36 |
sonne|work | some instructions I guess too | 19:37 |
@lisitsyn | sonne|work: what kind of instructions? | 19:37 |
sonne|work | and any user input | 19:37 |
@lisitsyn | sonne|work: say we don't send invitations, how to find c-base, etc? | 19:38 |
@lisitsyn | :D | 19:38 |
sonne|work | lisitsyn: I will send you something I received when registering for similar stuff | 19:38 |
@lisitsyn | sonne|work: okay thanks | 19:38 |
sonne|work | done | 19:39 |
@lisitsyn | sonne|work: so you are in france this weekend? | 19:39 |
@lisitsyn | err early next week | 19:40 |
@lisitsyn | :) | 19:40 |
sonne|work | yes | 19:40 |
sonne|work | taking vacations :D | 19:40 |
@lisitsyn | sonne|work: we should say we do not provide any accomodation and any invitation letters | 19:41 |
@lisitsyn | right? | 19:41 |
sonne|work | surely not but I dont' think it is necessary to say | 19:41 |
sonne|work | or? | 19:41 |
sonne|work | IDK | 19:42 |
@lisitsyn | sonne|work: I am afraid people might think | 19:42 |
@lisitsyn | about that | 19:42 |
sonne|work | maybe they need some info about hotels in berlin? | 19:42 |
@lisitsyn | sonne|work: I need | 19:42 |
sonne|work | but I don't think it is so difficult to find one | 19:42 |
@lisitsyn | :D | 19:42 |
sonne|work | google hotel berlin | 19:42 |
sonne|work | or hostel berlin | 19:42 |
sonne|work | done | 19:42 |
@lisitsyn | sonne|work: hah | 19:42 |
@lisitsyn | sonne|work: I called an agency today I hope they will find out something for me | 19:43 |
sonne|work | ohh cool sunsetting here | 19:43 |
@lisitsyn | sonne|work: Rungestrasse 20, 10179 Berlin, ???????? | 19:44 |
@lisitsyn | oops | 19:44 |
@lisitsyn | sonne|work: right? | 19:44 |
sonne|work | anything around train station jannowitzbruecke / ostbahnhof is close enough | 19:46 |
sonne|work | alright gtg | 19:46 |
sonne|work | cu | 19:46 |
@lisitsyn | see you | 19:46 |
wiking | mmmm | 19:48 |
wiking | weka's smo gives like 10% better accuracy | 19:48 |
wiking | than anything with shogun till now | 19:48 |
wiking | :( | 19:48 |
@lisitsyn | wiking: ehm! | 19:48 |
wiking | yes | 19:49 |
@lisitsyn | wiking: any guess? | 19:49 |
wiking | sad news | 19:49 |
wiking | really have no fucking clue why | 19:49 |
wiking | i'm now already trying kernel normalizers et | 19:49 |
@lisitsyn | wiking: same C? | 19:49 |
wiking | c | 19:49 |
@lisitsyn | wiking: you must be sure the C, the epsilon, etc are the same you know :) | 19:51 |
wiking | lisitsyn: c = 1.0 | 19:55 |
wiking | e = 1^-12 | 19:55 |
@lisitsyn | wiking: what is the kernel? | 19:55 |
wiking | polyt | 19:56 |
wiking | poly | 19:56 |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has quit [Quit: dvalcarce] | 19:58 | |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has joined #shogun | 20:02 | |
wiking | lisitsyn: any ideas? | 20:03 |
@lisitsyn | wiking: I'd compare kernel values to be sure | 20:04 |
wiking | lol | 20:10 |
wiking | this is crazy sad | 20:10 |
wiking | what shogun does here :) | 20:10 |
@lisitsyn | wiking: you've got to understand why :) | 20:13 |
wiking | lisitsyn: 93.053 % vs 0.820152 | 20:18 |
wiking | this is weka.smo vs shogun.libsvm | 20:19 |
wiking | and speed wise as well crazy faster in weka | 20:20 |
@lisitsyn | wiking: why do you compare accuracy btw? | 20:24 |
wiking | lisitsyn: because i'd like to have the same accuracy at least | 20:25 |
wiking | in shogun | 20:25 |
wiking | as i hate using weka | 20:25 |
wiking | :) | 20:25 |
@lisitsyn | wiking: I mean accuracy is not a good measure to compare | 20:26 |
wiking | should i print f1? | 20:26 |
wiking | :) | 20:26 |
@lisitsyn | wiking: no - but I'd have to ask you to find out if there is a bug :) | 20:26 |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has quit [Quit: dvalcarce] | 20:27 | |
wiking | lisitsyn: where? | 20:27 |
@lisitsyn | wiking: in libsvm probably | 20:28 |
wiking | lisitsyn: i've tried svmlight | 20:28 |
wiking | as well | 20:28 |
wiking | ;) | 20:28 |
@lisitsyn | wiking: but I'd check if C has the same meaning | 20:28 |
wiking | same shit | 20:28 |
@lisitsyn | wiking: then we either have a bug in apply (not train) or your parameters differ | 20:28 |
wiking | welll parameters are now the same | 20:29 |
wiking | even the cache size of the kernel | 20:29 |
wiking | :D | 20:29 |
@lisitsyn | wiking: normalization? standartization? | 20:31 |
wiking | lisitsyn: normalizing | 20:31 |
@lisitsyn | wiking: did you normalize it in shogun? | 20:31 |
wiking | hence i've 'fixed' in shogun PNorm | 20:31 |
wiking | :) | 20:31 |
wiking | this is why i had to template it | 20:32 |
wiking | ;P | 20:32 |
wiking | to be able to normalize the input | 20:32 |
@lisitsyn | wiking: why do you use pnorm? | 20:32 |
wiking | lisitsyn: what is better for normalizing? | 20:32 |
@lisitsyn | we have sumone and normone iirc? | 20:32 |
@lisitsyn | wiking: not better but what is P? | 20:33 |
wiking | lisitsyn: just checking now | 20:33 |
wiking | precision is 0.770022 | 20:33 |
wiking | on shogun | 20:33 |
wiking | 0.933 | 20:33 |
wiking | weka | 20:33 |
@lisitsyn | alright | 20:34 |
@lisitsyn | wiking: http://weka.sourceforge.net/doc/weka/classifiers/functions/SMO.html I see a lot of flags here | 20:34 |
@lisitsyn | I'd expect weka does some good stuff in normalizing something | 20:34 |
wiking | btw pnorm(2.0) = normone | 20:35 |
@lisitsyn | wiking: yes sure | 20:38 |
@lisitsyn | and pnorm(1.0) is sumone :) | 20:38 |
@lisitsyn | wiking: did you enable bias in libsvm? | 20:38 |
wiking | no | 20:40 |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has joined #shogun | 20:40 | |
@lisitsyn | wiking: should be enabled by default though | 20:40 |
wiking | i did this now | 20:40 |
wiking | svm->set_bias_enabled(true); | 20:40 |
wiking | just in case | 20:40 |
wiking | let's see if precision changes | 20:40 |
-!- RushabhM [~androirc@nat-128-84-124-0-240.cit.cornell.edu] has joined #shogun | 20:41 | |
wiking | lisitsyn: afaik it was enabled because precision is the exact same | 20:44 |
@lisitsyn | wiking: yes | 20:44 |
wiking | lol | 20:45 |
wiking | randomforest just nails it | 20:45 |
wiking | fuck | 20:45 |
wiking | there's nothing in shoung we can get near to these numbers on this data set | 20:45 |
wiking | there should be something done about it | 20:45 |
@lisitsyn | wiking: that's why you are here - no? :) | 20:45 |
-!- RushabhM [~androirc@nat-128-84-124-0-240.cit.cornell.edu] has quit [Ping timeout: 245 seconds] | 20:46 | |
-!- RushabhM [~androirc@nat-128-84-124-0-833.cit.cornell.edu] has joined #shogun | 20:48 | |
@lisitsyn | wiking: https://github.com/lisitsyn/tapkee/blob/master/include/tapkee_methods.hpp#L38 oops :D | 20:50 |
-!- RushabhM [~androirc@nat-128-84-124-0-833.cit.cornell.edu] has quit [Read error: Connection reset by peer] | 20:53 | |
-!- RushabhM [~androirc@nat-128-84-124-0-833.cit.cornell.edu] has joined #shogun | 20:54 | |
wiking | eheheh | 20:54 |
wiking | but really | 20:54 |
wiking | i will now implement randomforest into shogun | 20:54 |
@lisitsyn | wiking: there is no one that minds it :) | 20:55 |
wiking | hahaha yeah | 20:55 |
-!- RushabhM [~androirc@nat-128-84-124-0-833.cit.cornell.edu] has quit [Read error: Connection reset by peer] | 20:55 | |
wiking | i mind to do it | 20:56 |
-!- RushabhM [~androirc@nat-128-84-124-0-833.cit.cornell.edu] has joined #shogun | 20:56 | |
wiking | but i cannot fucking believe that we cannot do as good as weka | 20:56 |
@lisitsyn | wiking: there is no silver bullet - I guess RF doesn't work in kernel spaces :) | 20:56 |
@lisitsyn | so sonney2k hadn't to use it for his stuff | 20:57 |
wiking | but come on | 20:57 |
wiking | it's polykernel + smo | 20:58 |
wiking | we ought to have something that is near to what weka can offer | 20:58 |
@lisitsyn | wiking: I am somewhat sure it is a matter of preprocessing the data | 20:58 |
wiking | lisitsyn: yeah i thought so as well | 21:01 |
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:02 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 21:02 | |
-!- RushabhM [~androirc@nat-128-84-124-0-833.cit.cornell.edu] has quit [Remote host closed the connection] | 21:04 | |
dvalcarce | sonne|work: I've just finish demos integration in django, tomorrow i will check there are no bugs and send a pull request | 21:13 |
dvalcarce | good bye everyone :) | 21:13 |
@lisitsyn | bye | 21:13 |
@iglesiasg | bye bye | 21:14 |
-!- dvalcarce [~dvalcarce@86.117.117.91.dynamic.mundo-r.com] has left #shogun [] | 21:14 | |
@lisitsyn | iglesiasg: how is it going? | 21:14 |
@iglesiasg | lisitsyn: hey, all good here, what about you? | 21:14 |
@lisitsyn | fine | 21:15 |
@lisitsyn | iglesiasg: wiking is struggling with our svms now :) | 21:15 |
@iglesiasg | hehe with what exactly? | 21:15 |
@iglesiasg | in order to get something to work in travis? | 21:15 |
@lisitsyn | iglesiasg: no, weka works better than our svms | 21:16 |
@iglesiasg | lisitsyn: mmm we have space for improvement then | 21:16 |
@iglesiasg | in what sense to they work better btw? | 21:16 |
@lisitsyn | iglesiasg: the learned model appears to be better | 21:19 |
@iglesiasg | lisitsyn, wiking : have you tested with any particular dataset? | 21:19 |
-!- vgorbati_ [d4029f22@gateway/web/freenode/ip.212.2.159.34] has joined #shogun | 21:20 | |
-!- vgorbati [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 245 seconds] | 21:22 | |
-!- Asix3 [452f9490@gateway/web/freenode/ip.69.47.148.144] has quit [Ping timeout: 245 seconds] | 21:22 | |
wiking | iglesiasg: yes | 21:55 |
-!- vgorbati_ [d4029f22@gateway/web/freenode/ip.212.2.159.34] has quit [Ping timeout: 245 seconds] | 21:57 | |
@iglesiasg | wiking: so what did you try? | 22:14 |
@iglesiasg | I read something in logs about liblinear | 22:14 |
wiking | iglesiasg: i'm trying to do binary classification on a quite simple dataset | 22:20 |
wiking | but for some reason weka beats by a big margin any shogun classifier until now | 22:21 |
@iglesiasg | wiking: mmm not good | 22:23 |
wiking | mmm we don't have a possibility to standardize a dataset right? | 22:26 |
@lisitsyn | wiking: PruneVarSubMean | 22:27 |
@sonney2k | wiking, what do I hear weka works better? | 22:37 |
@sonney2k | what does that mean? | 22:37 |
@iglesiasg | this is a sad day for all of us :( | 22:38 |
@lisitsyn | haha | 22:38 |
@sonney2k | haha | 22:39 |
@sonney2k | iglesiasg, it is user error | 22:39 |
@iglesiasg | haha | 22:39 |
@sonney2k | no idea what wiking does but I am pretty sure that this is not a shogun bug | 22:39 |
@sonney2k | training with poly kernel needs a couple of tweaks to get the data behave well | 22:40 |
@sonney2k | the svm I mean | 22:40 |
@lisitsyn | sonney2k: what are the tricks? | 22:40 |
@sonney2k | there a lots of factors like kernel normalizer, or features normalization or inhomogeneous poly kernel or homogeneous... | 22:41 |
@sonney2k | everything can make a big difference | 22:41 |
@iglesiasg | lisitsyn: use weka *lol* | 22:41 |
@sonney2k | depending on the data | 22:41 |
@sonney2k | that weka trains a lot faster indicates that the data is better normalized | 22:41 |
wiking | sonney2k: yeah i'm trying with different normalization | 22:42 |
wiking | kernel and simple norm of the vectors | 22:42 |
@sonney2k | wiking, why do you use poly kernel? | 22:42 |
@sonney2k | why not gaussian? | 22:42 |
wiking | sonney2k: well i've tried different | 22:42 |
wiking | until now this is the best | 22:43 |
@sonney2k | gaussian is much easier to get good results with | 22:43 |
@sonney2k | wiking, just to understand - which parameters did you use for poly kernel? | 22:43 |
wiking | sonney2k: for polykernel itself i used the same cache size and degree as in weka | 22:44 |
wiking | degree only 1 | 22:44 |
@sonney2k | wiking, linear kernel? | 22:45 |
@lisitsyn | wait | 22:45 |
@lisitsyn | degree =1? | 22:45 |
@sonney2k | you mean with linear kernel you get good results? | 22:45 |
@sonney2k | why don't you use liblinear / svmocas then? | 22:46 |
@sonney2k | wiking, which kernel normalizer? | 22:46 |
wiking | sonney2k: not even near | 22:46 |
wiking | sonney2k: sqrtdiag | 22:46 |
@lisitsyn | wiking: what is the difference between linear and poly d=1? | 22:46 |
@sonney2k | for linear it doesn't make much sense | 22:46 |
@sonney2k | to put things on a hypersphere | 22:47 |
@sonney2k | lisitsyn, normalization | 22:47 |
@lisitsyn | sonney2k: ahh | 22:47 |
@sonney2k | otherwise nothing | 22:47 |
@lisitsyn | but I wouldn't go for that just because of normalization | 22:47 |
@sonney2k | wiking, do not compare accuracy - compare with something bias independent | 22:48 |
@sonney2k | like area under ROC curve | 22:48 |
wiking | sonney2k: lemme check | 22:48 |
@sonney2k | bias is very very sensitive | 22:49 |
@sonney2k | (might be...) | 22:49 |
@sonney2k | better plot both ROC curves even | 22:49 |
wiking | what is the the macro? | 22:49 |
* sonney2k leaving train | 22:49 | |
wiking | in contingencytable? | 22:49 |
@sonney2k | back in a few mins | 22:49 |
@lisitsyn | wiking: ROCEvaluation | 22:49 |
@lisitsyn | use it | 22:49 |
wiking | ah lol | 22:50 |
wiking | lol [ERROR] assertion m_current_values.vector && idx<get_num_labels() failed in file labels/Labels.cpp line 57 | 22:57 |
wiking | i wonder what's happening | 22:58 |
@lisitsyn | wiking: what did you do? | 22:58 |
wiking | just added CROCEvaluation roc; roc.evaluate(pred, gt); | 22:59 |
@lisitsyn | wiking: debug please :D | 23:00 |
@sonney2k | wiking, seems like your labels & predictions don't have the same size | 23:07 |
wiking | i wonder why dont we have such assert in contingencytable :S | 23:08 |
@lisitsyn | wiking: I am not sure it is the reason - did you check call stack? | 23:08 |
wiking | not yet | 23:08 |
-!- hoijui [~hoijui@dslb-088-074-124-102.pools.arcor-ip.net] has quit [Ping timeout: 252 seconds] | 23:11 | |
@sonney2k | wiking, so did you check now? | 23:11 |
@lisitsyn | sonney2k: http://d24w6bsrhbeh9d.cloudfront.net/photo/aM2oDxP_460s.jpg | 23:13 |
@lisitsyn | sonney2k: different lights in west/east berlin - really? | 23:14 |
@sonney2k | lisitsyn, you know the east :) | 23:22 |
@sonney2k | we can be lucky to even have electricity here | 23:23 |
@lisitsyn | sonney2k: hah why? | 23:23 |
-!- vikram360 [~vikram360@117.192.163.136] has quit [Ping timeout: 255 seconds] | 23:41 | |
--- Log closed Fri Apr 19 00:00:28 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!