--- Log opened Wed Mar 28 00:00:03 2012 | ||
--- Day changed Wed Mar 28 2012 | ||
blackburn | I am not shy with my alcohol addiction | 00:00 |
---|---|---|
blackburn | :D | 00:00 |
n4nd0 | sonney2k: I realized he gets funnier at nights :) | 00:00 |
@sonney2k | blackburn, tell n4nd0 about your bear! | 00:00 |
blackburn | oh its huge | 00:00 |
n4nd0 | what kind of bear are talking about? | 00:00 |
blackburn | my bear | 00:00 |
blackburn | he drinks a lot of vodka | 00:00 |
@sonney2k | the one from siberia | 00:00 |
blackburn | and eats people | 00:00 |
@sonney2k | kind of a pet blackburn has at home | 00:01 |
n4nd0 | you're kidding guys | 00:01 |
n4nd0 | I already tried that joke with polar bears with Spanish friends when I came to Sweden | 00:01 |
@sonney2k | I think he walks out his bear daily - scaring away people | 00:01 |
n4nd0 | :) | 00:01 |
blackburn | hah | 00:02 |
blackburn | wanna see my best photo? | 00:02 |
blackburn | http://cs579.userapi.com/u5562488/65536566/x_45297eef.jpg | 00:02 |
blackburn | LOL | 00:03 |
n4nd0 | haha wtf | 00:03 |
n4nd0 | it was just a set-up for the photo? | 00:03 |
blackburn | unfortunately no | 00:03 |
blackburn | :D | 00:03 |
blackburn | I should be ashamed of it probably but it is funny as hell | 00:04 |
@sonney2k | that's before blackburn is getting up - blood level too high (not enough alcohol-aka-fuel in blood yet) | 00:04 |
blackburn | actually it was new year celebrations | 00:05 |
blackburn | :D | 00:05 |
n4nd0 | no too bad for new years then | 00:06 |
blackburn | I've got one more funny photo from school | 00:06 |
blackburn | http://cs305.userapi.com/u18261492/43807744/x_942ecb5b.jpg | 00:06 |
blackburn | :D | 00:06 |
blackburn | n4nd0: this time yes - it is a set-up ;) | 00:07 |
n4nd0 | a Russian friend told me that people are more lucid when they drink ... then I started to believe stories about vodka :) | 00:07 |
n4nd0 | blackburn: good you said it :P | 00:07 |
n4nd0 | blackburn: do you guys have to go with suits to school? | 00:07 |
blackburn | but it is not a photoshop or anything | 00:07 |
blackburn | n4nd0: yes, we used to wear suits in school | 00:07 |
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has quit [Quit: Leaving] | 00:08 | |
blackburn | hmm I've got uncatchable bug | 00:15 |
blackburn | :D | 00:15 |
blackburn | n4nd0: sonney2k: good night guys | 00:19 |
@sonney2k | blackburn, do you know this https://code.google.com/p/c45/ | 00:19 |
blackburn | hmm | 00:19 |
blackburn | no | 00:19 |
@sonney2k | seems like someone wrote c45 in c++ | 00:19 |
@sonney2k | ...and is applying for gsoc | 00:19 |
@sonney2k | no idea how good his impl. is though... | 00:19 |
@sonney2k | but he could be the guy doing id3 / c45 / c50 / random forests | 00:20 |
blackburn | or girl according to email | 00:21 |
blackburn | ok see you | 00:21 |
-!- blackburn [~qdrgsm@188.168.128.91] has quit [Quit: Leaving.] | 00:21 | |
@sonney2k | night | 00:21 |
n4nd0 | good night | 00:21 |
n4nd0 | sonney2k: StreamingFeatures cannot be used directly by classifiers right? | 00:37 |
n4nd0 | in order to read from a file, I have to use StreamingFeatures to read the file and fill a e.g. CSimpleFeatures | 00:37 |
shogun-buildbot | build #400 of java_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/java_modular/builds/400 | 00:54 |
-!- jckrz [~jacek@89-69-164-5.dynamic.chello.pl] has quit [Ping timeout: 246 seconds] | 02:02 | |
-!- vikram360 [~vikram360@117.192.171.88] has joined #shogun | 02:22 | |
-!- vikram360 [~vikram360@117.192.171.88] has quit [Ping timeout: 246 seconds] | 02:30 | |
-!- vikram360 [~vikram360@117.192.170.214] has joined #shogun | 02:30 | |
-!- makokal_ [~makokal@brmn-4d0ae603.pool.mediaWays.net] has joined #shogun | 02:36 | |
-!- makokal [~makokal@brmn-4db71722.pool.mediaWays.net] has quit [Ping timeout: 245 seconds] | 02:38 | |
-!- makokal_ is now known as makokal | 02:38 | |
-!- flxb_ [~cronor@e178175102.adsl.alicedsl.de] has quit [Quit: flxb_] | 03:06 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 03:11 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Quit: flxb] | 03:16 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 03:43 | |
-!- pluskid [~chatzilla@60.191.2.230] has joined #shogun | 04:00 | |
pluskid | Finally figured out why I still cannot submit my application: I mistakenly registered as a mentor... | 04:01 |
pluskid | Writing to GSoC to fix this | 04:01 |
n4nd0 | pluskid: curious fact :) | 04:09 |
pluskid | n4nd0: I think maybe I clicked the wrong button when registering :( | 04:14 |
pluskid | Though I can simply register another account with another email :D | 04:14 |
n4nd0 | fast solucion :) | 04:15 |
n4nd0 | solution* | 04:15 |
pluskid | n4nd0: you are a GSoC, too, aren't you? | 04:16 |
pluskid | student | 04:16 |
n4nd0 | well, I try to be one | 04:16 |
pluskid | Cool, what idea are you working on? | 04:17 |
n4nd0 | SO | 04:17 |
n4nd0 | what about you? | 04:18 |
pluskid | what do you mean by SO? | 04:18 |
pluskid | I'm trying to work on structured output learning, is it the same topic to you? | 04:19 |
n4nd0 | oh sorry, Structured Output | 04:19 |
n4nd0 | yeah, it seems so :S | 04:19 |
pluskid | hmm... | 04:19 |
pluskid | there are two topics related to SO | 04:19 |
pluskid | Built generic structured output learning framework | 04:19 |
pluskid | and Bundle method solver for structured output learning | 04:20 |
pluskid | which one are you working on? | 04:20 |
n4nd0 | I am with the first one | 04:20 |
pluskid | OMG... | 04:20 |
n4nd0 | :S | 04:20 |
pluskid | You have been very active here, I should catch up... >_< | 04:21 |
n4nd0 | well it looks like you have quite a bit of experience in ml (info in your presentation mail) it's me who should catch up actually ;) | 04:24 |
-!- vikram360 [~vikram360@117.192.170.214] has quit [Ping timeout: 252 seconds] | 04:25 | |
zxtx | wasn't SO on the list last year? | 04:26 |
n4nd0 | no idea | 04:27 |
pluskid | no idea, too. Haven't participated last year. | 04:28 |
pluskid | n4nd0: Yes, both of us have our own strength. But I think I should start to consider applying one more project/organization for the sake of conflict. | 04:30 |
zxtx | http://www.shogun-toolbox.org/gsoc/ideas-2011.html#so | 04:30 |
n4nd0 | pluskid: probably, I think I will apply for a second project in shogun as well | 04:30 |
pluskid | me too, because shogun is really an ideal project to work with. | 04:31 |
pluskid | maybe we should discuss before we choose our second idea | 04:31 |
pluskid | to avoid a 2nd conflict :D | 04:31 |
n4nd0 | haha | 04:32 |
n4nd0 | I already thought of generic multiclass | 04:32 |
n4nd0 | what about you? | 04:32 |
pluskid | haven't decided yet | 04:32 |
pluskid | maybe I will consider other ideas, since you chose MTL | 04:33 |
pluskid | I'm still reading the details about other ideas | 04:33 |
n4nd0 | ok | 04:33 |
pluskid | I will tell you when I have a decision | 04:33 |
n4nd0 | pluskid: cool, thanks :) | 04:34 |
-!- makokal [~makokal@brmn-4d0ae603.pool.mediaWays.net] has quit [Quit: makokal] | 04:38 | |
-!- shelhamer [~shelhamer@c-24-62-205-17.hsd1.ma.comcast.net] has joined #shogun | 04:38 | |
-!- shelhamer [~shelhamer@c-24-62-205-17.hsd1.ma.comcast.net] has quit [Client Quit] | 04:43 | |
pluskid | n4nd0: Maybe I'll choose MTL.... but another MTL: Implement multitask and domain adaptation algorithms | 04:43 |
pluskid | or Kernel Two-sample/Dependence test | 04:44 |
n4nd0 | ok | 04:44 |
n4nd0 | I think someone else is with the multitask | 04:44 |
n4nd0 | but it's rather difficult to avoid competence | 04:45 |
pluskid | yes :( | 04:45 |
zxtx | does anyone ever need multitask learning in practice? | 04:45 |
pluskid | it seems useful, because many tasks alone have too few information to work out a good model. | 04:48 |
pluskid | But I don't know about the real-world application side | 04:48 |
zxtx | well it seems you just need a good prior across tasks | 05:02 |
zxtx | anyway clearly not a project I should be a mentor or student on :P | 05:03 |
pluskid | zxtx: yes, but I guess in MTL the piror can be learned automatically from the data, instead of made by the expert :p | 05:06 |
-!- pluskid [~chatzilla@60.191.2.230] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120312181643]] | 05:19 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 260 seconds] | 05:29 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has left #shogun ["Leaving"] | 05:37 | |
-!- harshit_ [~harshit@182.68.160.94] has joined #shogun | 06:31 | |
-!- pluskid [~chatzilla@60.191.2.230] has joined #shogun | 07:12 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 07:28 | |
-!- pluskid [~chatzilla@60.191.2.230] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120312181643]] | 07:33 | |
harshit_ | hey sonney2k | 07:38 |
harshit_ | hey n4nd0 | 07:38 |
n4nd0 | hey harshit_ how is it going? | 07:39 |
harshit_ | good . whats up on your side ? | 07:39 |
n4nd0 | just fine :) | 07:39 |
harshit_ | so i read the irclog | 07:39 |
harshit_ | and i think my deep learning thing could not work in this gsoc | 07:40 |
n4nd0 | ok | 07:40 |
n4nd0 | did you take a look to those references sonney2k talked about? | 07:40 |
harshit_ | but still i can implement it after gsoc ( or if i dont get into gsoc) | 07:40 |
harshit_ | yeah | 07:40 |
n4nd0 | ok, then it's good | 07:41 |
harshit_ | they are basically big toolboxes | 07:41 |
harshit_ | based on Neural networks | 07:41 |
harshit_ | so extracting deep learning algos from them is a real pain | 07:41 |
n4nd0 | I see | 07:41 |
harshit_ | and you were also talking about implementing decision trees | 07:42 |
n4nd0 | yes | 07:42 |
harshit_ | C5.0 code seems to be easy, but i wonder why blackburn kept on saying | 07:42 |
n4nd0 | it looks like shogun needs them very bad | 07:43 |
harshit_ | that implementing that is really difficult | 07:43 |
harshit_ | oh i se | 07:43 |
harshit_ | see* | 07:43 |
harshit_ | but if C5.0 code is in C then what is so difficult in implementing it in shogun ? | 07:44 |
harshit_ | do you have any idea | 07:44 |
n4nd0 | well probably it is not so difficult as you have said | 07:45 |
n4nd0 | but it is not going to be probably just copy + paste | 07:45 |
n4nd0 | one probably has to adapt it to shogun's structure (e.g. using SGVectors, SGMatrices and the like) | 07:45 |
n4nd0 | I have not really looked into that project though | 07:46 |
harshit_ | okay.. | 07:46 |
harshit_ | i'll have a talk with blackburn though, | 07:46 |
harshit_ | Coz after implementing newtonsvm my project seems to be small | 07:46 |
n4nd0 | I think it's a good idea | 07:47 |
harshit_ | i need something else to add to that project | 07:47 |
harshit_ | do you know of anything that could be added to it !! | 07:47 |
n4nd0 | there must be other tree implementations that could be useful to have | 07:47 |
n4nd0 | didn't they mention something about it? I think I have heard it | 07:48 |
harshit_ | other tree implementations ?? as in | 07:49 |
harshit_ | you talking about c4.5 or other implementation of c5.0 itself | 07:49 |
n4nd0 | ahm | 07:49 |
harshit_ | i think blackburn will give me a closer idea.. | 07:50 |
harshit_ | and some trivial questions: | 07:50 |
harshit_ | what are these shogun-builtbot and shogun-toolbox in chat area | 07:51 |
harshit_ | and how is sonney2k always available | 07:51 |
n4nd0 | shogun-buildbot just gives some log information about the state of the build of the project | 07:52 |
n4nd0 | you do a pull request for example, sonney2k merges it | 07:52 |
n4nd0 | and a automatic process of building the project starts | 07:52 |
harshit_ | okay got it | 07:52 |
n4nd0 | this guy shogun-buildbot gives us feedback on IRC about how did that process go | 07:53 |
n4nd0 | checked some of the links it gives if you want | 07:53 |
n4nd0 | I have no idea what shogun-toolbox does | 07:53 |
harshit_ | and second question how is sonney2k always available | 07:53 |
harshit_ | yeah i checked once.. | 07:53 |
n4nd0 | and well sonney2k may have an IRC client connected all the time | 07:54 |
n4nd0 | but that doesn't mean he is online all the time | 07:54 |
harshit_ | yeah.. | 07:54 |
n4nd0 | you should ask him about it if you are curious :) | 07:54 |
harshit_ | i would i want to do it with my account too :) | 07:55 |
-!- Marty28 [~chatzilla@158.181.76.57] has joined #shogun | 08:25 | |
-!- vikram360 [~vikram360@117.192.170.214] has joined #shogun | 08:32 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 08:35 | |
Marty28 | hiho | 08:35 |
-!- hpsoar__ [7c108901@gateway/web/freenode/ip.124.16.137.1] has joined #shogun | 08:37 | |
n4nd0 | Marty28: hey! | 08:38 |
Marty28 | Could anyone please tell me if the toolbox is ready for Python 3.1? | 08:38 |
blackburn | Marty28: almost, gsomix works on it actualyl | 08:38 |
blackburn | actually* | 08:38 |
Marty28 | Ok, so I will better go on with 2.7 for the next months. | 08:39 |
Marty28 | Is there a random DNA/RNA sequence generator with motifs included in shogun somewhere? | 08:42 |
blackburn | I haven't seen any :) | 08:44 |
Marty28 | I am currently programming one, but I want it to be realistic. | 08:44 |
-!- jckrz [~jacek@89-69-164-5.dynamic.chello.pl] has joined #shogun | 08:45 | |
harshit_ | hey blackburn | 08:47 |
blackburn | harshit_: hi | 08:47 |
Marty28 | hiho | 08:47 |
harshit_ | sonney told me not to use get_computed_dot_feature_matrix() function | 08:48 |
harshit_ | and use add_to_dense() etc | 08:48 |
blackburn | yes makes more sense | 08:48 |
harshit_ | but the problem is that i want that matrix for cblas_dgemm | 08:48 |
harshit_ | so is there any way out ? | 08:49 |
blackburn | what is the matrix you compute? | 08:49 |
harshit_ | i dont compute any matrix | 08:49 |
harshit_ | its just the feature matrix entered by user | 08:49 |
blackburn | what is the matrix you compute with dgemm? | 08:49 |
harshit_ | actually it has been used at alot of places | 08:50 |
harshit_ | in one place i have to get a subset of feature matrix | 08:50 |
blackburn | gram or covariance? | 08:50 |
harshit_ | with given rows | 08:50 |
harshit_ | gram , covariance ?? i dont get it | 08:51 |
blackburn | e.g. X'X is a gram matrix | 08:51 |
blackburn | and XX' is a covariance matrix | 08:51 |
blackburn | where X is a feature matrix | 08:51 |
harshit_ | ohh | 08:51 |
harshit_ | gram | 08:51 |
harshit_ | then | 08:51 |
blackburn | you may employ linear kernel for that | 08:52 |
blackburn | or compute it manually | 08:53 |
blackburn | while (i,j) slot contains just a dot product between ith and jth vectors | 08:53 |
harshit_ | okay.. got it | 08:54 |
harshit_ | but that would require complete redesign of whole program | 08:54 |
blackburn | why? | 08:55 |
harshit_ | I dont get that why cant we make use of whole feature matrix | 08:55 |
harshit_ | does that slows down the program ? | 08:55 |
blackburn | no, but whole feature matrix could be huge | 08:56 |
-!- hpsoar__ [7c108901@gateway/web/freenode/ip.124.16.137.1] has quit [Quit: Page closed] | 08:56 | |
blackburn | and wouldn't fit memory | 08:56 |
harshit_ | okay.. then i think its worth working on that. | 08:57 |
-!- jckrz [~jacek@89-69-164-5.dynamic.chello.pl] has quit [] | 08:58 | |
blackburn | main point is that you should avoid this access operations and use add and dot ops when possible | 09:00 |
-!- hpsoar [7c108901@gateway/web/freenode/ip.124.16.137.1] has joined #shogun | 09:01 | |
harshit_ | yeah got it , Bcoz thats the reason why COFFIN framework was built . | 09:03 |
harshit_ | hey blackburn i am using CDotFeatures for features but in this class add_to_dense and dense_dot are virtual, so do i need to use some other type of features in my program | 09:17 |
blackburn | no, it is ok | 09:17 |
blackburn | it is an abstract class for other features | 09:17 |
blackburn | so e.g. SimpleFeatures has implementation of these methods | 09:18 |
harshit_ | but then i could not make use of add_to_dense once i type cast to cdotfeatures | 09:18 |
blackburn | why? | 09:18 |
harshit_ | bcoz it wont knw which implementation to use | 09:19 |
harshit_ | i am not sure actually | 09:19 |
blackburn | you don't have to know :) | 09:19 |
harshit_ | okay ... i'll try it and get back to you in a moment. | 09:20 |
n4nd0 | blackburn: do you have any example reading from .mat like we do with loadmat in python examples but in C++? | 09:31 |
n4nd0 | blackburn: I am doing one using Streaming objects but I think I am not getting it well | 09:31 |
blackburn | n4nd0: oh.. no, I never tried to do it with C++ | 09:33 |
blackburn | why do you want to work with .mat from C++? | 09:33 |
n4nd0 | I just want to use data files from C++ | 09:34 |
blackburn | http://scikit-learn.org/dev/modules/clustering.html#overview-of-clustering-methods | 09:34 |
n4nd0 | they don't have any specific format, just one line with two floats | 09:34 |
n4nd0 | should I use shogun stuff to read them or standard C++? | 09:34 |
n4nd0 | nice plot | 09:35 |
harshit_ | hey, what is dense vector in shogun ?? i mean why is dense word used in every function in Cdotfeatures | 09:36 |
n4nd0 | harshit_: is just a normal matrix, non sparse | 09:36 |
harshit_ | matrix ?? | 09:37 |
blackburn | vector matrix all the same :) | 09:37 |
harshit_ | okay | 09:37 |
harshit_ | :) | 09:37 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 09:51 | |
-!- Marty28 is now known as Marty | 09:57 | |
-!- Marty is now known as Marty28 | 09:57 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 10:24 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 10:44 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 10:46 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 10:51 | |
-!- gsomix [~gsomix@188.168.13.36] has quit [Read error: Operation timed out] | 10:53 | |
-!- flxb_ [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 10:57 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Ping timeout: 260 seconds] | 10:59 | |
-!- flxb_ is now known as flxb | 10:59 | |
Marty28 | ah, easysvm has a generator for DNA with motifs | 11:05 |
-!- harshit_ [~harshit@182.68.160.94] has quit [Ping timeout: 252 seconds] | 11:18 | |
-!- hpsoar [7c108901@gateway/web/freenode/ip.124.16.137.1] has quit [Quit: Page closed] | 11:59 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Quit: flxb] | 12:10 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 12:11 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Client Quit] | 12:11 | |
-!- OgreMagi [~vinaygarg@14.139.82.6] has joined #shogun | 12:13 | |
-!- OgreMagi [~vinaygarg@14.139.82.6] has left #shogun [] | 12:13 | |
-!- gsomix [~gsomix@188.168.5.159] has joined #shogun | 12:23 | |
gsomix | hi | 12:23 |
-!- harshit_ [~harshit@182.68.160.94] has joined #shogun | 12:26 | |
-!- pluskid [~chatzilla@60.191.2.230] has joined #shogun | 12:27 | |
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has joined #shogun | 12:30 | |
harshit_ | hey, does anyone here know the working of add_to_dense_vec() ?? | 12:41 |
harshit_ | problem is that it gives me the results, considering features matrix as row major but actually feature matrix is in column major | 12:42 |
-!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 12:46 | |
-!- gsomix [~gsomix@188.168.5.159] has quit [Ping timeout: 244 seconds] | 13:13 | |
-!- PSmitAalto [82e9b263@gateway/web/freenode/ip.130.233.178.99] has joined #shogun | 13:25 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 13:29 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Client Quit] | 13:29 | |
-!- gsomix [~gsomix@83.234.54.41] has joined #shogun | 13:29 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 246 seconds] | 13:32 | |
-!- in3xes [~in3xes@210.212.58.168] has joined #shogun | 13:45 | |
PSmitAalto | Hey guys | 14:07 |
PSmitAalto | A question about using MKLMultiClass, will this use CPLEX if I have compiled with CPLEX support? | 14:07 |
-!- in3xes_ [~in3xes@180.149.49.230] has joined #shogun | 14:14 | |
-!- in3xes [~in3xes@210.212.58.168] has quit [Ping timeout: 260 seconds] | 14:18 | |
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has quit [Ping timeout: 265 seconds] | 14:21 | |
-!- in3xes_ [~in3xes@180.149.49.230] has quit [Ping timeout: 246 seconds] | 14:33 | |
-!- in3xes_ [~in3xes@180.149.49.230] has joined #shogun | 14:46 | |
-!- in3xes__ [~in3xes@180.149.49.230] has joined #shogun | 14:51 | |
-!- in3xes_ [~in3xes@180.149.49.230] has quit [Ping timeout: 244 seconds] | 14:54 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 14:55 | |
-!- hpsoar_ [7c108901@gateway/web/freenode/ip.124.16.137.1] has joined #shogun | 15:04 | |
-!- vikram360 [~vikram360@117.192.170.214] has quit [Ping timeout: 264 seconds] | 15:07 | |
hpsoar_ | I'm wondering why there's no comments or suggestions to my proposal for gsoc | 15:07 |
hpsoar_ | I suppose that someone may review it and give some judgement and suggestions | 15:08 |
-!- vikram360 [~vikram360@117.192.170.214] has joined #shogun | 15:19 | |
-!- in3xes__ is now known as in3xes | 15:29 | |
-!- in3xes [~in3xes@180.149.49.230] has quit [Quit: Leaving] | 15:29 | |
-!- hpsoar_ [7c108901@gateway/web/freenode/ip.124.16.137.1] has quit [Quit: Page closed] | 15:34 | |
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has joined #shogun | 15:45 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 15:55 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 15:59 | |
-!- vikram360 [~vikram360@117.192.170.214] has quit [Ping timeout: 244 seconds] | 16:06 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 16:24 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Client Quit] | 16:24 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 16:24 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 16:40 | |
harshit_ | hey n4nd0 | 17:10 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 17:10 | |
harshit_ | i hav an issue with add_to_dense_vec() function | 17:10 |
harshit_ | hav you used it anywhere ? | 17:10 |
-!- vikram360 [~vikram360@117.192.170.214] has joined #shogun | 17:12 | |
n4nd0 | harshit_: I think I have not used it, but tell me anyway | 17:14 |
harshit_ | firstly i need to confirm that whether matrices are stored in shogun in column major order or not | 17:17 |
harshit_ | i think they are stored in column major .. yes | 17:17 |
harshit_ | ? | 17:17 |
n4nd0 | yes they are | 17:18 |
harshit_ | problem is that add_to_dense_vec gives me the results, considering features matrix is stored as row major | 17:18 |
n4nd0 | mmm | 17:18 |
n4nd0 | so you are using CDotFeatures? | 17:19 |
harshit_ | do you knw any place where i can upload small scripts ? | 17:19 |
harshit_ | yes sort of coz simplefeatures is derived from CDotFeatures only | 17:19 |
n4nd0 | if you mean code I use this http://snipt.org/ | 17:20 |
harshit_ | nice site, have a look at a small portion of my code :http://snipt.org/uggF9 | 17:21 |
harshit_ | here when i use CMath::display_matrix(sgv.vector,x_d,1); it shows me 1 2 3 | 17:23 |
harshit_ | rather than 1 4 7 | 17:23 |
harshit_ | same is the case with add_to_dense_vec | 17:23 |
n4nd0 | let me think of it a moment | 17:24 |
harshit_ | ok | 17:24 |
n4nd0 | so sgv is the first vector of features | 17:26 |
n4nd0 | i.e. the first column of matrix | 17:26 |
n4nd0 | so then I think it's normal that it's 1 2 3 | 17:26 |
harshit_ | it should be first row of matrix ! | 17:27 |
n4nd0 | why the first row? | 17:28 |
n4nd0 | as we said they are stored column wise in shogun | 17:28 |
harshit_ | but still if then if it first vector (or first coloumn ) then it should contain x_n elements | 17:29 |
n4nd0 | and here x_n is? | 17:29 |
n4nd0 | what are the values of x_n and x_d here? | 17:29 |
harshit_ | x_n: no of features .. ie no rows | 17:30 |
harshit_ | and x_d is no of dimensions | 17:30 |
n4nd0 | yeah ... but I mean the particular values you are using | 17:30 |
harshit_ | see matrix is [1 4 7;2 5 8 ;3 6 9] | 17:31 |
harshit_ | ohh its 5 for x_n and 3 for x_d | 17:31 |
n4nd0 | in any case I think you got it the other way round | 17:31 |
n4nd0 | look at the constructor in CSimpleFeatures http://www.shogun-toolbox.org/doc/en/current/SimpleFeatures_8cpp_source.html#l00032 | 17:32 |
n4nd0 | the one in line 32 | 17:32 |
n4nd0 | num_feat (your x_d here) is for the rows | 17:32 |
n4nd0 | and num_vec (your x_n) for the number of cols | 17:32 |
n4nd0 | got it? | 17:34 |
harshit_ | but i have used set_feature_matrix | 17:34 |
harshit_ | and in set_feature_matrix() first argument is num_feat and second is num_vec | 17:34 |
harshit_ | and i have used x_d for first | 17:34 |
harshit_ | and x_n for second | 17:35 |
n4nd0 | yeah but why did you make that? | 17:36 |
harshit_ | just for understanding this function | 17:36 |
n4nd0 | I mean | 17:36 |
-!- jckrz [~jrzeniewi@89-69-164-5.dynamic.chello.pl] has joined #shogun | 17:36 | |
n4nd0 | SGMatrix<float64_t> matrix(x_n,x_d); | 17:36 |
n4nd0 | first x_n and later x_d | 17:37 |
n4nd0 | why | 17:37 |
n4nd0 | features->set_feature_matrix(matrix.matrix,x_d,x_n); | 17:37 |
n4nd0 | the other way round | 17:37 |
harshit_ | coz SGMatrix constructor says to have first arg as no of rows and second as no of columns | 17:37 |
harshit_ | if i do other way around i get an assert error | 17:38 |
harshit_ | hey u figured out something | 17:38 |
n4nd0 | I don't know if you see what happens | 17:38 |
n4nd0 | it is like if you have this matrix | 17:38 |
harshit_ | may be i should not use SGMatrix | 17:39 |
n4nd0 | [1 3 5; 2 4 6] | 17:39 |
n4nd0 | and you build it doing SGMatrix< float64_t >(2, 3) | 17:39 |
harshit_ | wait i'll give it a try | 17:39 |
n4nd0 | and later you do features(matrix, 3, 2) | 17:39 |
n4nd0 | you will tell to features that the matrix is | 17:40 |
harshit_ | in first case ie with the matrix you wrote | 17:41 |
harshit_ | sgv vector displays 1 2 3 | 17:41 |
n4nd0 | exactly] | 17:42 |
harshit_ | whereas it should display 1 3 5 | 17:42 |
harshit_ | how | 17:42 |
n4nd0 | no | 17:42 |
n4nd0 | each feature vector is a column | 17:43 |
harshit_ | okay agreed | 17:43 |
n4nd0 | I know that the notation might be confusing | 17:43 |
harshit_ | but then column should have 2 members | 17:44 |
harshit_ | take a practical example | 17:44 |
n4nd0 | I don't know what you mean with but then column should have 2 members | 17:44 |
n4nd0 | ok I think I got it | 17:45 |
n4nd0 | yeah it should be like that | 17:45 |
n4nd0 | but the problem is when you do features->set_feature_matrix(matrix.matrix,x_d,x_n); | 17:45 |
n4nd0 | changing the order | 17:45 |
-!- pluskid [~chatzilla@60.191.2.230] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120312181643]] | 17:45 | |
n4nd0 | you create it with x_n, x_d | 17:45 |
n4nd0 | and set it with x_d, x_n | 17:45 |
n4nd0 | when your matrix really means | 17:46 |
n4nd0 | [1 3 5; 2 4 6] | 17:46 |
n4nd0 | you make believe to features that it is | 17:46 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 17:46 | |
harshit_ | but doing that results in an error | 17:46 |
n4nd0 | I guess the error must come from another thing that is wrong | 17:47 |
harshit_ | i think best way to debug this is to drop SGMatrix for now and feed simplefeatures normal matrix | 17:47 |
n4nd0 | mmm I don't think so | 17:47 |
n4nd0 | when you change to x_d, x_n | 17:48 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Client Quit] | 17:48 | |
n4nd0 | everything must work well until SGMatrix<float64_t> sgmatrix=features->get_computed_dot_feature_matrix(); | 17:48 |
n4nd0 | right? | 17:49 |
harshit_ | wait let me use gdb | 17:49 |
harshit_ | it is good till add_to_dense_vec () function | 17:51 |
harshit_ | and sgv displays : 1 2 | 17:51 |
harshit_ | and matrix is= [1 4; 2 5 ; 3 6] | 17:52 |
-!- PSmitAalto [82e9b263@gateway/web/freenode/ip.130.233.178.99] has quit [Ping timeout: 245 seconds] | 17:53 | |
harshit_ | dam this is really confusing | 17:54 |
harshit_ | i think i need to have a closer look at how other implementations make use of add_to_dense_vec() function | 17:56 |
n4nd0 | I think you've to think how it is stored in memory | 18:00 |
n4nd0 | is just an array | 18:00 |
n4nd0 | a vector is an array a matrix is also an array | 18:00 |
harshit_ | yeah right | 18:00 |
n4nd0 | for matrices the dimensions are the ones that define them, and are really important | 18:00 |
harshit_ | but see here dimensions are alright i mean whe i use get_dim_feature_space it returns 3 | 18:02 |
harshit_ | exactly the value i used while defining matrix | 18:02 |
harshit_ | and when i use add_to_dense_vec it also returns 3 values | 18:02 |
harshit_ | but it returns those 3 values considering the matrix as row major | 18:03 |
harshit_ | but actually it is stored in column major in features | 18:03 |
harshit_ | i think we need blackburn around here for this .. | 18:04 |
harshit_ | or perhaps sonney2k | 18:05 |
harshit_ | i think i am messing up with x_n and x_d .. | 18:05 |
harshit_ | function could not be wrong .. | 18:06 |
-!- Miggy [~piggy@14.139.82.6] has joined #shogun | 18:07 | |
harshit_ | Hey dont waste your time on this (if you are ), i'll ask blackburn or sonney2k for this | 18:09 |
n4nd0 | there is no problem | 18:09 |
n4nd0 | but I don't really understand what is it that you don't understand now :) | 18:10 |
n4nd0 | harshit_: what is it that doesn't make sense for you? | 18:11 |
harshit_ | If add_to_dense_vec is returning 3 elements i.e no. of dimensions i.e one of feature vector | 18:11 |
harshit_ | then it should be 1 4 7 | 18:12 |
harshit_ | first datapoint | 18:12 |
n4nd0 | your matrix was [1 4 7;2 5 8 ;3 6 9] | 18:12 |
n4nd0 | ?? | 18:12 |
harshit_ | yes | 18:12 |
n4nd0 | your ; is a row delimiter right? | 18:13 |
n4nd0 | I mean | 18:13 |
n4nd0 | 1 4 7 | 18:13 |
n4nd0 | 2 5 8 | 18:13 |
harshit_ | yes | 18:13 |
n4nd0 | ok | 18:13 |
n4nd0 | the first point is 1 2 3 | 18:13 |
harshit_ | exactly | 18:13 |
n4nd0 | so then? | 18:14 |
harshit_ | oh sorry | 18:14 |
harshit_ | ohkay | 18:14 |
harshit_ | okay | 18:14 |
harshit_ | okay | 18:14 |
harshit_ | got it | 18:14 |
n4nd0 | :D | 18:14 |
n4nd0 | the key is to remember that they are just arrays | 18:15 |
n4nd0 | just a T* (T is float64_t, int32_t or whatever) | 18:15 |
harshit_ | i was confusing coz i thought 1 4 7 as my first datapoint always | 18:16 |
n4nd0 | ok | 18:17 |
harshit_ | thanks | 18:18 |
n4nd0 | you're welcome | 18:19 |
n4nd0 | gtg | 18:26 |
n4nd0 | bye | 18:26 |
-!- vikram360 [~vikram360@117.192.170.214] has quit [Read error: Connection reset by peer] | 18:31 | |
harshit_ | bye n4nd0 you really provide great help | 18:31 |
-!- vikram360 [~vikram360@117.192.170.214] has joined #shogun | 18:32 | |
n4nd0 | harshit_: :) | 18:33 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 18:33 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 18:35 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 18:48 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 18:57 | |
-!- harshit_ [~harshit@182.68.160.94] has quit [Quit: Leaving] | 19:11 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has joined #shogun | 19:15 | |
CIA-64 | shogun: Evgeniy Andreev master * rda550bd / (3 files in 3 dirs): fixed python typemaps - http://git.io/nVB85w | 20:06 |
CIA-64 | shogun: Soeren Sonnenburg master * r355cb51 / (3 files in 3 dirs): | 20:06 |
CIA-64 | shogun: Merge pull request #399 from gsomix/python3_interface | 20:06 |
CIA-64 | shogun: updates on python3 interface - http://git.io/cFomOQ | 20:06 |
CIA-64 | shogun: Soeren Sonnenburg master * rc69fb70 / (3 files): | 20:12 |
CIA-64 | shogun: Merge pull request #398 from karlnapf/master | 20:12 |
CIA-64 | shogun: First bunch of fixes for migration - http://git.io/baZr9g | 20:12 |
gsomix | sonney2k, moin. | 20:20 |
@sonney2k | gsomix, hey | 20:23 |
@sonney2k | dobrui wetscher :) | 20:24 |
gsomix | sonney2k, dobrui. :) | 20:24 |
-!- blackburn [~qdrgsm@188.168.2.79] has joined #shogun | 20:25 | |
@sonney2k | gsomix, I wanted to comment on your patch: | 20:25 |
gsomix | I'm listening. | 20:26 |
@sonney2k | python3's integers are long? | 20:26 |
@sonney2k | and python3's strings are always unicode? | 20:27 |
@sonney2k | and python3 supports some kind of bytestrings that are mutable? | 20:27 |
@sonney2k | if so we should for strings use these mutable byte based strings | 20:27 |
@sonney2k | (at least in the future support it) | 20:28 |
@sonney2k | blackburn, hi hi | 20:29 |
blackburn | sonney2k: heyaa | 20:29 |
gsomix | sonney2k, integers are long. And strings are unicode, but there is old-style strings - PyBytes. | 20:29 |
@sonney2k | gsomix, are these PyBytes deprecated or the new mutable string type? | 20:30 |
blackburn | oh that is so funny they call everything with py- suffix | 20:30 |
blackburn | pytoilet | 20:30 |
@sonney2k | blackburn, 8 proposals by now... | 20:31 |
@sonney2k | blackburn, I am happy they do - C has no namespaces... | 20:31 |
@sonney2k | and actually namespaces are crank ... | 20:31 |
@sonney2k | when a project doesn't use them one is screwed | 20:31 |
blackburn | sonney2k: yeah that's ok but every library has this py-* | 20:32 |
blackburn | I said suffix? :D | 20:32 |
blackburn | prefix probably lol | 20:32 |
@sonney2k | SG*crap | 20:32 |
gsomix | sonney2k, new. PyBytes has just replaced PyString. | 20:33 |
blackburn | sonney2k: I heard you like proposals so n4ndo put a proposal into your melange | 20:33 |
@sonney2k | gsomix, so is PyBytes immutable or mutable? | 20:33 |
@sonney2k | blackburn, I know | 20:34 |
blackburn | I am in mood to joke - again damn | 20:34 |
blackburn | sonney2k: so what about C4.5 girl? | 20:34 |
gsomix | sonney2k, immutable. | 20:35 |
@sonney2k | I don't know - the google code project is very new (didn't exist a few days ago) | 20:35 |
@sonney2k | and we have to see how it goes to integrate that stuff in shogun / compare with j45 from weka | 20:36 |
blackburn | sonney2k: actually we really need to know # of slots.. | 20:36 |
gsomix | sonney2k, hmm. wait. | 20:38 |
gsomix | sonney2k, yes, bytes is immutable. I'm confused by old documents. | 20:41 |
-!- lamer13329602986 [~jrzeniewi@89-69-164-5.dynamic.chello.pl] has joined #shogun | 20:45 | |
* gsomix had returned from dance floor. Now he is bit stupid. | 20:46 | |
blackburn | lol | 20:46 |
-!- jckrz [~jrzeniewi@89-69-164-5.dynamic.chello.pl] has quit [Ping timeout: 265 seconds] | 20:48 | |
* blackburn is getting angry with girl-like laugh of his neighbor :D | 20:50 | |
@sonney2k | gsomix, so b'string' is a byte string | 20:50 |
gsomix | sonney2k, yep | 20:50 |
@sonney2k | and 'string' is a normal string right? | 20:50 |
gsomix | unicode | 20:51 |
blackburn | no more u'stuff'? | 20:51 |
blackburn | ugliest part of python for sure | 20:51 |
@sonney2k | gsomix, b'string' works in python 2.x too right? | 20:51 |
@sonney2k | blackburn, yeah luckily no longer needed in python3 | 20:51 |
gsomix | sonney2k, it works. In my python2.7. :) | 20:52 |
gsomix | But type(b'string') returns 'str' in python2.7. | 20:53 |
@sonney2k | gsomix, ok I guess we have to support both the b'string' variant and the unicode variant (checking that only 0...127 is used) in shogun | 20:53 |
@sonney2k | not now but we should keep that in mind | 20:54 |
gsomix | sonney2k, ok. | 20:55 |
gsomix | I remember it. | 20:55 |
gsomix | sonney2k, can i send changes in examples? | 20:56 |
gsomix | print ' ' -> print(''), and others. | 20:57 |
@sonney2k | sure | 20:57 |
@sonney2k | these are all compatible so it is a good idea | 20:58 |
-!- flxb_ [~cronor@e178172254.adsl.alicedsl.de] has joined #shogun | 20:59 | |
gsomix | sonney2k, and what about examples, which are not working at now? | 21:00 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 21:00 | |
* gsomix is enslaved by google translate. The thing was invented by the devil. =___= | 21:01 | |
-!- flxb [~cronor@e178175102.adsl.alicedsl.de] has quit [Ping timeout: 246 seconds] | 21:01 | |
-!- flxb_ is now known as flxb | 21:01 | |
blackburn | huh | 21:01 |
@sonney2k | hi wiking - are you still with us? torturing alexander with questions? | 21:01 |
wiking | sonney2k: yeye | 21:01 |
wiking | i'm here | 21:02 |
blackburn | sonney2k: hmm actually I forgot to CC you probably | 21:02 |
wiking | just got back | 21:02 |
wiking | blackburn: yeah actually we could do an abstract class | 21:02 |
wiking | and that would be the 'minimizer' | 21:02 |
wiking | so that one has to implement (inherit) a class from that one | 21:02 |
wiking | did you mean that in your email? | 21:02 |
blackburn | yes | 21:03 |
blackburn | better this way | 21:03 |
wiking | ok that's good as well | 21:03 |
wiking | more c++ like | 21:03 |
wiking | i can change now the api in 10 mins | 21:03 |
@sonney2k | gsomix, difficult to say | 21:03 |
wiking | any other comments? | 21:03 |
wiking | blackburn: the LatentLabel would actually contain two parts | 21:04 |
blackburn | wiking: well only ones I have listed | 21:04 |
wiking | y (the usual labeling) and h and that can be 'anyting' | 21:04 |
wiking | that is the latent variable | 21:04 |
wiking | and how you interpret that | 21:04 |
wiking | it's really up to your application | 21:04 |
@sonney2k | gsomix, quite a couple are not working but it needs probably quite some work to figure out why ... | 21:04 |
wiking | sonney2k: yeah i did a first api attempt for the latent svm.... do you want to be in the flow of emails? | 21:05 |
gsomix | sonney2k, ok, i will think about it. | 21:05 |
@sonney2k | wiking, I would even prefer if you do this discussion in public on the mailinglist | 21:05 |
wiking | sonney2k: actually why haven't we started there i have no idea :)))) | 21:06 |
wiking | i second this idea as well :P | 21:06 |
gsomix | sonney2k, while I will solve the problems in the covertree. | 21:06 |
blackburn | sonney2k: wiking: hmm could be considered unfair | 21:06 |
wiking | the only thing is that i have a fierce competition ... :) | 21:06 |
wiking | somebody else wants as well to do the latent svm | 21:06 |
wiking | i've just seen a week ago on the mailing list | 21:07 |
blackburn | it would look like we already chose wiking | 21:07 |
wiking | but then again i do all my implementation public | 21:07 |
@sonney2k | blackburn, no | 21:07 |
@sonney2k | last year a couple of people started doing some project | 21:07 |
wiking | so that could be taken an improved by anybody... | 21:07 |
@sonney2k | some we took some we didn't take | 21:08 |
@sonney2k | and we even merged most patches | 21:08 |
blackburn | wiking: sonney2k: ok then it should be in the way 'how would I do it if I was accepted' :D | 21:08 |
@sonney2k | if someone wants to do a project - why should he waste time submitting patches he is not really interested in | 21:09 |
@sonney2k | so he could start directly with the project and send patches for that | 21:09 |
@sonney2k | e.g. harshit is doing newtonsvm for some time now | 21:09 |
wiking | sonney2k: i would send patches... just simply referring to my fork's branch.... | 21:09 |
wiking | ;) | 21:09 |
@sonney2k | pull requests whatever | 21:10 |
wiking | hihi yeah sure... anyhow then i'll send the next email about it on the mailing list | 21:10 |
blackburn | wiking: yeah just make work in progress request | 21:11 |
blackburn | all the discussion could be there | 21:11 |
@sonney2k | last year we had people working on shogun i/o stuff, sgd-qn, java interface already before gsoc started | 21:14 |
blackburn | really? | 21:14 |
blackburn | :D | 21:14 |
blackburn | I forgot | 21:14 |
@sonney2k | blackburn, I have no idea what you did? | 21:15 |
@sonney2k | vodka? | 21:15 |
blackburn | yeah | 21:15 |
blackburn | a lot of vodking | 21:15 |
@sonney2k | blackburn, lets call you vodking from now on :D | 21:15 |
blackburn | yeah | 21:16 |
wiking | :D | 21:16 |
blackburn | sonney2k: evaluation stuff | 21:16 |
blackburn | I reworked it a little (completely) | 21:16 |
CIA-64 | shogun: Soeren Sonnenburg master * r6fea6c9 / src/interfaces/python_modular/swig_typemaps.i : remove debug code / string typemap - http://git.io/r9qi1g | 21:17 |
@sonney2k | vodking - I know and GNB, conjugateindex... | 21:18 |
blackburn | sonney2k: do not remind meconjugateindex crap | 21:18 |
blackburn | it was developed in december actually | 21:18 |
blackburn | but GNB during pre-gsoc yeah | 21:18 |
@sonney2k | and actually you were mentoring students | 21:18 |
blackburn | hahah | 21:18 |
@sonney2k | btw, is GNB aready converted to logspace? | 21:19 |
@sonney2k | would be some simple task for a patch | 21:19 |
blackburn | yes | 21:19 |
blackburn | unfortunately | 21:19 |
@sonney2k | ahh | 21:19 |
@sonney2k | hmmhhh | 21:19 |
blackburn | but covertree for KNN is still vacant | 21:19 |
blackburn | sonney2k: you said you were planning to try GNB on some task, did you? | 21:20 |
@sonney2k | really many examples work now :) | 21:20 |
@sonney2k | err tests | 21:20 |
@sonney2k | s/examples/tests/ | 21:20 |
blackburn | sonney2k: which fails? | 21:24 |
@sonney2k | quite a couple of errors we had before | 21:25 |
@sonney2k | some nan's in substracts due to preprocessors | 21:25 |
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has quit [Ping timeout: 248 seconds] | 21:26 | |
-!- Peter_Gr [~Peter_Gr@115.248.130.148] has joined #shogun | 21:28 | |
CIA-64 | shogun: Soeren Sonnenburg master * rcfb5860 / testsuite/python_modular/tester.py : add -f option to tester to only show failed tests - http://git.io/ZIRqqg | 21:33 |
-!- vikram360 [~vikram360@117.192.170.214] has quit [Ping timeout: 246 seconds] | 21:33 | |
@sonney2k | gsomix, btw could you please use assert_version to check for the python version | 21:43 |
@sonney2k | gsomix, in configure - an example is assert_version doxygen $_doxygen $_doxygen_version 1.6.0 | 21:44 |
-!- vikram360 [~vikram360@117.192.170.214] has joined #shogun | 21:44 | |
@sonney2k | _doxygen contains yes/no/auto (depending on value it will fail) | 21:44 |
@sonney2k | _doxygen_version is the version string and 1.6.0 is what is required | 21:44 |
@sonney2k | blackburn, what's up with converters - will they ever get a test? | 21:53 |
blackburn | sonney2k: yes | 21:53 |
blackburn | but tests are rather complex for that shit | 21:54 |
shogun-buildbot | build #594 of cmdline_static is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/594 blamelist: heiko.strathmann@gmail.com | 21:54 |
blackburn | sonney2k: arpack+superlu / arpack / lapack | 21:55 |
blackburn | three eigensolvers | 21:55 |
blackburn | while weight matrices are computed with pretty high precision and could be checked | 21:56 |
blackburn | eigenvectors are pretty random | 21:56 |
@sonney2k | some kind of sanity check - like objective value? | 21:56 |
blackburn | great idea btw | 21:56 |
blackburn | but it could vary as well | 21:57 |
blackburn | I'll check later | 21:57 |
@sonney2k | (I wouldn't accept any mloss paper that doesn't have tests btw) | 21:57 |
-!- vikram360 [~vikram360@117.192.170.214] has quit [Ping timeout: 264 seconds] | 21:57 | |
@sonney2k | but I am no longer AE so you might be lucky | 21:57 |
blackburn | haha | 21:57 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 21:58 | |
@sonney2k | some of the tests in shogun are maybe overly strict | 21:58 |
shogun-buildbot | build #572 of r_static is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/572 blamelist: heiko.strathmann@gmail.com | 21:58 |
blackburn | sonney2k: well do you think it is ok to resubmit in say may? | 21:58 |
@sonney2k | currently I expect the objects to be exactly the same | 21:58 |
blackburn | is there any time limit for that? | 21:58 |
@sonney2k | I am not aware of a limit | 21:59 |
blackburn | I can't promise I implement all the tests even in april | 21:59 |
@sonney2k | that makes sense if you test on the same machine with no difference in anything (numerical libraries, compiler, ...) | 21:59 |
blackburn | yes | 21:59 |
@sonney2k | but for e.g. svm just using atlas will make the tests fail | 21:59 |
@sonney2k | and the solution will only be within 1e-3 or whatever svm eps is | 22:00 |
blackburn | sonney2k: may be test should contain required precision | 22:00 |
@sonney2k | problem is I often introduced problems that were like 1e-10 off somewhere but results should have been *exactly* the same | 22:01 |
@sonney2k | blackburn, yeah but this can hide problems too | 22:01 |
@sonney2k | maybe we should generate these test data on the buildbot | 22:01 |
blackburn | sonney2k: one reason why I didn't implement such test is it is really easy to check graphically | 22:01 |
@sonney2k | and then be more tolerant on other machines | 22:02 |
shogun-buildbot | build #573 of octave_static is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/573 blamelist: heiko.strathmann@gmail.com | 22:02 |
@sonney2k | blackburn, I don't agree. a PCA solution might *look* ok but variance of that other dimension might be 1e-6 larger... | 22:02 |
@sonney2k | or swiss-unroll might look almost perfect but still isn ot | 22:03 |
@sonney2k | is not | 22:03 |
blackburn | sonney2k: I meant LLE and such stuff | 22:03 |
blackburn | actually there is no degree of freedom in lle | 22:03 |
blackburn | variance one* | 22:03 |
@sonney2k | blackburn, what do you think wrt my test idea above? | 22:04 |
blackburn | sonney2k: no idea if it would work well | 22:04 |
@sonney2k | blackburn, it will work well until we change build bot / os / gcc / libs | 22:05 |
blackburn | hmm | 22:05 |
@sonney2k | and on other archs we can at least test... | 22:05 |
blackburn | you are right probably | 22:05 |
@sonney2k | so we need some precision for returned outputs in these functions | 22:05 |
shogun-buildbot | build #561 of python_static is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/561 blamelist: heiko.strathmann@gmail.com | 22:06 |
@sonney2k | either pairs ((output1, prec2), (output2, prec2)) | 22:06 |
@sonney2k | or (out1,out2), (prec1,prec2) | 22:06 |
@sonney2k | or some global precisions variable | 22:06 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 22:07 | |
blackburn | sonney2k: actually I can employ some student for writing tests for edrt :D | 22:07 |
blackburn | objective function would be provided so it is pretty easy to do | 22:07 |
blackburn | in python | 22:07 |
@sonney2k | sounds like you are a lot faster doing it yourself | 22:08 |
blackburn | I don't know whether it should be implemented in edrt itself | 22:08 |
blackburn | what do you think? | 22:08 |
blackburn | sonney2k: ok I can't promise still but I'll try to add testing there | 22:10 |
blackburn | as soon as I am finished with libedrt | 22:10 |
blackburn | sonney2k: good thing is LLE/LTSA/KLTSA/KLLE/NPE/LLTSA are all the same | 22:10 |
blackburn | so 6 methods are almost ready | 22:11 |
@sonney2k | ...waste of time | 22:11 |
blackburn | wrong! | 22:11 |
blackburn | you will see it is not | 22:11 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 22:12 | |
blackburn | sonney2k: there was no way to embed 100K vectors before | 22:12 |
blackburn | it would take a hour but it would fit to memory this time | 22:12 |
blackburn | it was all matrix based | 22:12 |
blackburn | now it is more sparse and compact | 22:13 |
blackburn | yes I could do that in the way it was done before but it would take *more* time because of not flexible design | 22:13 |
blackburn | I am 500% sure with it | 22:13 |
-!- romi__ [~mizobe@189-18-40-233.dsl.telesp.net.br] has joined #shogun | 22:14 | |
@sonney2k | blackburn, but you don't need that weird libedrt / compute callbacks to do that - that really is a different issue | 22:15 |
blackburn | sonney2k: yes exactly what I need to make it more flexible | 22:15 |
blackburn | sonney2k: the only alternative is a class structure with some virtual compute kernel methods | 22:16 |
blackburn | sonney2k: it was kernel matrix based | 22:16 |
@sonney2k | but why was it kernel matrix based before? kernel(i,j) did always exist | 22:17 |
blackburn | sonney2k: because of wrong design! | 22:17 |
blackburn | and this way fixes it | 22:18 |
@sonney2k | I thought you did this because it was easier to call whatever inversion/eigenvalue lapack functions | 22:18 |
blackburn | sonney2k: well things changed a great deal there :D | 22:18 |
blackburn | sonney2k: now there is less code and more flexibility | 22:19 |
-!- Marty28 [~chatzilla@158.181.76.57] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120310010446]] | 22:19 | |
blackburn | for new methods / complex cachine one wouldn't have to create new kernel | 22:19 |
blackburn | caching* | 22:19 |
@sonney2k | kernel caching should be in CKernel anyways | 22:20 |
blackburn | ok someday I'll convince you | 22:21 |
blackburn | ?? | 22:21 |
blackburn | :D | 22:21 |
@sonney2k | not this century :D | 22:22 |
blackburn | ok then next one | 22:22 |
blackburn | :) | 22:22 |
blackburn | I have to go | 22:22 |
blackburn | see you tomorrow | 22:22 |
-!- blackburn [~qdrgsm@188.168.2.79] has quit [Quit: Leaving.] | 22:23 | |
@sonney2k | n4nd0, very nice application... | 22:24 |
shogun-buildbot | build #192 of nightly_none is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/192 | 22:24 |
@sonney2k | n4nd0, and thanks for helping harshit | 22:24 |
@sonney2k | I hope he manages to use DotFeatures... | 22:24 |
gsomix | sonney2k, ok. | 22:25 |
CIA-64 | shogun: Soeren Sonnenburg master * r0ff05d8 / testsuite/python_modular/tester.py : show case / exception in tester - http://git.io/uLZKGg | 22:27 |
CIA-64 | shogun: Soeren Sonnenburg master * r9d9bf6e / src/shogun/features/BinnedDotFeatures.cpp : initializ m_features with NULL / fixes hang - http://git.io/7lQflA | 22:27 |
@sonney2k | n4nd0, ahh could you please file a bug - such that we remember to implement SGNDArray serialization support? | 22:28 |
n4nd0 | sonney2k: hey! | 22:31 |
n4nd0 | sonney2k: so where can I file that bug? | 22:31 |
@sonney2k | n4nd0, github -> create new issue | 22:32 |
n4nd0 | sonney2k: ok, I am going to do that now | 22:32 |
n4nd0 | sonney2k: apart from that, I have been using this StreamingFile and StreamingSimpleFeatures classes | 22:33 |
@sonney2k | and? | 22:33 |
n4nd0 | sonney2k: I am detecting some memory leaks with valgrind | 22:33 |
n4nd0 | sonney2k: do you know if they are supposed to be completely ok? | 22:33 |
n4nd0 | sonney2k: or could it be the case that they leak? | 22:33 |
@sonney2k | they should be yes | 22:33 |
@sonney2k | if not -> bug | 22:33 |
n4nd0 | sonney2k: ok, I will take it a closer look before doing it | 22:34 |
n4nd0 | sonney2k: about the application, something I should complete or change then? | 22:34 |
-!- romi_ [~mizobe@189-18-40-233.dsl.telesp.net.br] has quit [Quit: Leaving] | 22:45 | |
shogun-buildbot | build #595 of cmdline_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/595 | 22:45 |
-!- gsomix [~gsomix@83.234.54.41] has quit [Read error: Operation timed out] | 22:46 | |
@sonney2k | yay! of the 313 test configurations 'just' 49 fail | 22:49 |
shogun-buildbot | build #573 of r_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/573 | 22:50 |
n4nd0 | :) | 22:51 |
shogun-buildbot | build #574 of octave_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/574 | 22:55 |
shogun-buildbot | build #562 of python_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/562 | 22:59 |
-!- gsomix [~gsomix@188.168.2.195] has joined #shogun | 23:02 | |
-!- lamer13329602986 [~jrzeniewi@89-69-164-5.dynamic.chello.pl] has quit [Quit: jIRCii - http://www.oldschoolirc.com] | 23:10 | |
gsomix | sonney2k, can you help me? | 23:13 |
@sonney2k | gsomix, ? | 23:13 |
gsomix | sonney2k, http://pastebin.com/3neDVhcZ | 23:14 |
gsomix | I always have problems with git repository. =___= | 23:15 |
@sonney2k | what do you want to do? | 23:15 |
@sonney2k | e.g. to drop your changes you could do git checkout -f src/shogun/base/Parameter.cpp | 23:16 |
@sonney2k | and then the pull | 23:16 |
gsomix | sonney2k, I just want to get the latest changes. | 23:17 |
@sonney2k | gsomix, yeah but you changed sth in files someone else changed so now you have a conflict | 23:17 |
@sonney2k | question is what you want now | 23:17 |
@sonney2k | get their version or keep yours | 23:18 |
@sonney2k | or manually resolve changes | 23:18 |
@sonney2k | for example you can specify git pull -s STRATEGY | 23:19 |
@sonney2k | where STRATEGY is their or our | 23:19 |
@sonney2k | theirs / ours | 23:19 |
@sonney2k | so the remote / your version will be used | 23:20 |
@sonney2k | or you edit the files to resolve the conflict | 23:21 |
@sonney2k | and then do git add FILE_WITH_CONFLICT | 23:21 |
@sonney2k | and git commit when done | 23:21 |
gsomix | sonney2k, but, I did not change these files. | 23:23 |
gsomix | Hmm. | 23:23 |
shogun-buildbot | build #424 of octave_modular is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_modular/builds/424 blamelist: heiko.strathmann@gmail.com | 23:24 |
@sonney2k | I would say you did | 23:25 |
@sonney2k | so do the git rebase master -s theirs | 23:25 |
@sonney2k | if you are sure | 23:25 |
gsomix | sonney2k, thanks. | 23:46 |
shogun-buildbot | build #399 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/399 | 23:53 |
-!- OgreMagi [~vinaygarg@14.139.82.6] has joined #shogun | 23:57 | |
OgreMagi | hi | 23:58 |
OgreMagi | @sonney2k: I want to apply for the project " Built Generic Multiclass Learning Framework ". How should I proceed and are there any potential candidates for this particular project ? | 23:59 |
--- Log closed Thu Mar 29 00:00:19 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!