--- Log opened Wed Mar 21 00:00:19 2012 | ||
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 00:23 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 00:42 | |
-!- in3xes [~in3xes@210.212.58.111] has quit [Ping timeout: 246 seconds] | 00:46 | |
-!- l0nr4n [~l0nr4n@unaffiliated/l0nr4n] has quit [Quit: This computer (or maybe me) has gone to sleep] | 02:32 | |
-!- GenX [~Sambhav@14.139.82.6] has quit [Ping timeout: 246 seconds] | 02:42 | |
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds] | 03:28 | |
-!- GenX [~Sambhav@14.139.82.6] has joined #shogun | 03:33 | |
-!- GenX__ [~Sambhav@14.139.82.6] has joined #shogun | 03:41 | |
-!- GenX [~Sambhav@14.139.82.6] has quit [Ping timeout: 250 seconds] | 03:44 | |
-!- Sambhav__ [~Sambhav@14.139.82.6] has joined #shogun | 03:55 | |
-!- GenX__ [~Sambhav@14.139.82.6] has quit [Ping timeout: 252 seconds] | 03:58 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 04:02 | |
-!- Sambhav__ [~Sambhav@14.139.82.6] has quit [Ping timeout: 246 seconds] | 04:13 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 265 seconds] | 04:14 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 04:42 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 260 seconds] | 04:48 | |
-!- in3xes [~in3xes@59.163.196.121] has joined #shogun | 05:45 | |
-!- in3xes [~in3xes@59.163.196.121] has quit [Ping timeout: 244 seconds] | 05:50 | |
CIA-64 | shogun: Sergey Lisitsyn master * rdf1b927 / (2 files): Updated set_weights method - http://git.io/tyV-SA | 06:55 |
---|---|---|
CIA-64 | shogun: Sergey Lisitsyn master * r6b510fc / src/interfaces/modular/Library.i : Restored plif array things - http://git.io/a9QZiA | 06:55 |
CIA-64 | shogun: Sergey Lisitsyn master * rebe175c / (6 files in 3 dirs): Fixes for applications - http://git.io/hkYWEg | 06:55 |
CIA-64 | shogun: Sergey Lisitsyn master * r0e3adb0 / .gitignore : Updated git ignore - http://git.io/mmq9Rg | 06:55 |
-!- GenX [~Sambhav@14.139.82.6] has joined #shogun | 07:45 | |
-!- GenX [~Sambhav@14.139.82.6] has quit [Ping timeout: 245 seconds] | 08:01 | |
CIA-64 | shogun: Soeren Sonnenburg master * rba918e6 / (src/Makefile src/Makefile.template): add ctags target to upper makefile - http://git.io/iYJ41A | 08:10 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 08:13 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 08:30 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 08:48 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 09:03 | |
-!- GenX [~Sambhav@14.139.82.6] has joined #shogun | 09:37 | |
n4nd0 | sonne|work: hi! I am thinking of applying for the project "various usability improvements" apart from SO to raise the possibilities of success a bit | 09:42 |
n4nd0 | sonne|work: I have not previous coding experience with swig though, do you think that should prevent me from applying for that project then? | 09:42 |
-!- Miggy [~piggy@14.139.82.6] has joined #shogun | 10:26 | |
-!- Miggy [~piggy@14.139.82.6] has quit [Ping timeout: 240 seconds] | 10:30 | |
-!- l0nr4n [~l0nr4n@hugogascon.ml.tu-berlin.de] has joined #shogun | 10:52 | |
-!- l0nr4n [~l0nr4n@hugogascon.ml.tu-berlin.de] has quit [Changing host] | 10:52 | |
-!- l0nr4n [~l0nr4n@unaffiliated/l0nr4n] has joined #shogun | 10:52 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 11:09 | |
-!- puneetgoyal [~chatzilla@115.240.0.222] has joined #shogun | 11:14 | |
-!- harshit_ [~harshit@182.68.142.173] has joined #shogun | 11:14 | |
harshit_ | hello everyone | 11:17 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 11:17 | |
n4nd0 | hi | 11:18 |
harshit_ | @n4nd0 are you in the team of shogun ? | 11:20 |
n4nd0 | harshit_: I have made some contributions to the code, but I am just an student | 11:22 |
harshit_ | okay, Just had a few questions related to shogun's framework can you help me out | 11:23 |
n4nd0 | of course, as long as I have the answers :P | 11:24 |
harshit_ | if i want to do some operations on labels can i use CMath functions on it ? | 11:26 |
-!- GenX [~Sambhav@14.139.82.6] has quit [Ping timeout: 250 seconds] | 11:27 | |
n4nd0 | harshit_: so CLabels have a member that is a SGVector called labels | 11:28 |
harshit_ | operations as in scalar multiplication, or dot multiplication with any other vector | 11:28 |
n4nd0 | exactly | 11:28 |
n4nd0 | probably you can use that member to do the operations | 11:28 |
harshit_ | that class : SGVector contains all the functions for these opns within it ? | 11:29 |
n4nd0 | that class contains some functions but is not very exhaustive | 11:29 |
harshit_ | wait i'll hav a look at its signature | 11:30 |
n4nd0 | however, it has a public member | 11:30 |
n4nd0 | T * where T is likely to be int32_t or float64_t | 11:30 |
n4nd0 | so you can use CMath operations on that member | 11:30 |
harshit_ | okay thanks bro. | 11:31 |
harshit_ | and also i sent a mail on shogun's mailing list, But didnt get any reply | 11:31 |
harshit_ | if you knw some thing related to those questions | 11:31 |
harshit_ | please help me out, I am currently stuck in middle | 11:31 |
n4nd0 | I read your mail yesterday | 11:32 |
n4nd0 | but I couldn't answer most of the stuff | 11:32 |
n4nd0 | let me read again and refresh, maybe we can discuss about it anyway | 11:32 |
n4nd0 | so what does matlab minres exactly do? | 11:33 |
harshit_ | Thanks, In advance | 11:33 |
harshit_ | it calculates minimum resolution of a matrix | 11:33 |
harshit_ | i also dont knw much abt it, Maybe | 11:33 |
harshit_ | some linear algebra guy will have more knowledge abt it | 11:34 |
n4nd0 | I don't know what is the minimum resolution of a matrix :P | 11:34 |
harshit_ | and second question: hav you come across any function in shogun for issparse() | 11:34 |
n4nd0 | so for the issparse thing | 11:35 |
n4nd0 | I know that in shogun there are structures prepared to hold dense data like CDotFeatures | 11:35 |
n4nd0 | and sparse, CSparseFeatures | 11:35 |
n4nd0 | so what is exactly the use that they make in the code of issparse? | 11:35 |
n4nd0 | is it just to test if the structure is itself sth sparse? | 11:36 |
n4nd0 | or do they actually check the content of the matrix to check whether it should be store as sparse? | 11:36 |
n4nd0 | stored* | 11:36 |
harshit_ | it checks the matrix if it is sparse or dense | 11:37 |
n4nd0 | mmm | 11:37 |
n4nd0 | but do you know how they make this check? | 11:37 |
harshit_ | exactly, Same was my reaction when i came across this function | 11:38 |
harshit_ | i can only say one thing MATLAB is gr8, it provides so complex thing in so much easy form | 11:38 |
n4nd0 | take a look here | 11:38 |
n4nd0 | http://snipt.org/ugxc2 | 11:39 |
n4nd0 | is from the matlab doc | 11:39 |
n4nd0 | that makes me believe that they just check the type of container | 11:39 |
n4nd0 | let me see if I get to the actual code | 11:39 |
harshit_ | thats weird ..no ? | 11:40 |
n4nd0 | why do you think so? | 11:41 |
harshit_ | how can they decide whether it is sparse or not on basis of class being used ? | 11:42 |
n4nd0 | I think matlab can store matrices in two (at least) different ways | 11:43 |
n4nd0 | dense representation and sparse | 11:44 |
n4nd0 | if you don't know what I mean, do like this in matlab | 11:44 |
n4nd0 | A = [1 2; 3 4] | 11:44 |
n4nd0 | sparse(A) | 11:44 |
n4nd0 | you'll notice the different representation | 11:44 |
n4nd0 | so I guess that internally they have something to check if the matrix is stored in one or the other way | 11:45 |
harshit_ | yeah | 11:45 |
n4nd0 | probably that issparse is just checking the value of that "something" | 11:45 |
harshit_ | what value ? | 11:46 |
harshit_ | i didnt get that last point | 11:46 |
harshit_ | i think they must have some algo | 11:46 |
n4nd0 | the value of the type of object | 11:46 |
harshit_ | to check whether to store the matrix in sparse or dense format | 11:46 |
n4nd0 | not whether TO store but whether IT is stored in sparse or dense | 11:47 |
n4nd0 | but again, I'm not sure about this | 11:47 |
n4nd0 | is just how I understand it from the doc of the file I pasted before | 11:48 |
harshit_ | but we dont make any explicit call to store our matrices in dense or sparse format while creating it ! | 11:48 |
n4nd0 | what do you mean with our? in shogun? | 11:48 |
harshit_ | no in matlab . when we create any matrix | 11:49 |
n4nd0 | well | 11:49 |
n4nd0 | if you just define it like A = [....] | 11:49 |
n4nd0 | it will be dense | 11:49 |
n4nd0 | if you do sparse(A) | 11:49 |
n4nd0 | the return value is sparse | 11:49 |
n4nd0 | check it with | 11:49 |
n4nd0 | issparse(A) ( returns 0 ) | 11:50 |
n4nd0 | issparse( sparse(A) ) ( returns 1 ) | 11:50 |
harshit_ | yeah i saw that | 11:50 |
harshit_ | , you made a good point with it | 11:50 |
n4nd0 | :) | 11:50 |
harshit_ | thanks i'll explore more about it | 11:50 |
n4nd0 | I'm not sure about it though, so check it always just in case | 11:50 |
harshit_ | yeah i know.. :) | 11:51 |
n4nd0 | in order to translate this to shogun I'd say that | 11:51 |
n4nd0 | if you deal with CFeatures in your function | 11:51 |
n4nd0 | you should check if it is CDotFeatures or CSparseFeatures | 11:51 |
n4nd0 | I am checking minres now | 11:52 |
n4nd0 | it is for the minimum residual, I have heard that before at least :) | 11:52 |
harshit_ | I think that wont be right .. but i'll hav a look on it | 11:52 |
harshit_ | oh yeah, i wrote it wrong last time | 11:53 |
n4nd0 | probably there is a method in blas for the minimum residual | 11:53 |
harshit_ | but also min res is of special type .. | 11:56 |
n4nd0 | ?? | 11:56 |
harshit_ | its Hessian | 11:57 |
harshit_ | i dont knw what it is | 11:57 |
harshit_ | but it is given in the comment with minres() in .m file | 11:57 |
n4nd0 | the second derivative of a matrix, like the Jacobian but one order more | 11:57 |
n4nd0 | mmmmm | 11:58 |
n4nd0 | but I guess that if the function in matlab computes the minimum residual | 11:58 |
n4nd0 | a function that does the same should work | 11:58 |
harshit_ | is that Hessian you are talkin about | 11:58 |
n4nd0 | yes | 11:59 |
harshit_ | yeah i'll have a look at clbas function list | 11:59 |
harshit_ | i could not find minres in cblas . | 12:01 |
harshit_ | checked all level 1,2,3 | 12:01 |
harshit_ | btw why are cblas functions divided into level 1,2,3 | 12:02 |
harshit_ | do you hav any idea ? | 12:02 |
n4nd0 | opeartions between vectors, vectors and matrices and matrices; that's for the levels | 12:02 |
harshit_ | okay, so what i am looking for should be in level 3 | 12:03 |
harshit_ | but there is nothing like that | 12:03 |
harshit_ | are you there / | 12:07 |
n4nd0 | yes | 12:07 |
n4nd0 | but I don't know how to compute that either so I am trying to find sth on the web | 12:07 |
harshit_ | okay , thanks for your time | 12:08 |
n4nd0 | for me it looks like this function you do the job | 12:11 |
n4nd0 | http://www.netlib.org/templates/double/GMRES.f | 12:11 |
n4nd0 | it's a LAPACK funtion so I'd say it can be used from shogun | 12:12 |
harshit_ | i dont think whole lapack is included in shogun | 12:13 |
harshit_ | is it? | 12:13 |
harshit_ | coz last time i searched shogun's CMath library i didn't find any thing like this | 12:14 |
n4nd0 | things in CMath are things that are implemented in shogun | 12:14 |
n4nd0 | but shogun uses external libraries and packages | 12:14 |
n4nd0 | so I think you can use that functions from shogun including the appropriate header | 12:15 |
harshit_ | and while linking do i need to include appropriate library file and library path too ? | 12:16 |
harshit_ | or every thing is in -lshogun itself | 12:16 |
n4nd0 | what do you normally do to compile? | 12:17 |
n4nd0 | I just run make and I think that with that it should be enough | 12:17 |
harshit_ | okay i'll have a look at shogun's Makefile, That will probably tell me what all external libraries are included | 12:18 |
n4nd0 | ok | 12:19 |
harshit_ | damm shogun's makefile is really huge i havnt dealt with any thing like that before | 12:21 |
harshit_ | and the function that you just told me gmres.f | 12:27 |
harshit_ | it is written in fortran | 12:27 |
harshit_ | how can i use it in C++ | 12:27 |
n4nd0 | in lapack.cpp you can see how other lapack functions are used from shogun | 12:29 |
harshit_ | okay ,thanks. | 12:29 |
harshit_ | got it, Thanks for your help | 12:31 |
n4nd0 | no problem | 12:31 |
n4nd0 | have to go now | 12:31 |
n4nd0 | byew | 12:31 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 12:31 | |
-!- harshit_ [~harshit@182.68.142.173] has quit [Quit: Leaving] | 12:32 | |
-!- puneetgoyal_ [~chatzilla@115.242.49.216] has joined #shogun | 12:33 | |
-!- puneetgoyal [~chatzilla@115.240.0.222] has quit [Ping timeout: 252 seconds] | 12:35 | |
-!- puneetgoyal_ is now known as puneetgoyal | 12:35 | |
-!- PhilTillet [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 12:40 | |
-!- gsomix [~gsomix@85.26.233.219] has joined #shogun | 12:47 | |
gsomix | hi | 12:48 |
-!- blackburn [~qdrgsm@83.234.54.68] has joined #shogun | 12:55 | |
-!- blackburn [~qdrgsm@83.234.54.68] has quit [Client Quit] | 12:56 | |
-!- puneetgoyal [~chatzilla@115.242.49.216] has quit [Remote host closed the connection] | 12:59 | |
-!- blackburn [~qdrgsm@83.234.54.68] has joined #shogun | 13:01 | |
gsomix | blackburn, yo | 13:01 |
blackburn | gsomix: this time feel free to contact me via jabber :) | 13:02 |
blackburn | gsomix: or personally | 13:02 |
gsomix | blackburn, 5 minute | 13:05 |
-!- gsomix [~gsomix@85.26.233.219] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 13:06 | |
-!- puneetgoyal [~puneet@115.242.20.205] has joined #shogun | 13:16 | |
-!- menonnik [~chatzilla@59.178.167.103] has joined #shogun | 13:51 | |
-!- mayanks43 [~Marcus@115.248.130.148] has joined #shogun | 13:52 | |
-!- blackburn [~qdrgsm@83.234.54.68] has quit [Ping timeout: 272 seconds] | 14:16 | |
-!- blackburn [~qdrgsm@83.234.54.68] has joined #shogun | 14:16 | |
-!- PhilTillet [~Philippe@tillet-p42154.maisel.int-evry.fr] has quit [Ping timeout: 272 seconds] | 14:38 | |
-!- gsomix [~gsomix@85.26.232.52] has joined #shogun | 15:01 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 15:05 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 15:08 | |
-!- in3xes [~in3xes@59.163.196.121] has joined #shogun | 15:09 | |
-!- vikram360 [~vikram360@117.192.187.94] has joined #shogun | 15:10 | |
-!- gsomix [~gsomix@85.26.232.52] has quit [Ping timeout: 246 seconds] | 15:33 | |
blackburn | sonney2k: sonne|work: are any of you guys around? ;) | 15:35 |
-!- gsomix [~gsomix@83.149.21.22] has joined #shogun | 15:45 | |
-!- in3xes [~in3xes@59.163.196.121] has quit [Ping timeout: 246 seconds] | 16:08 | |
-!- menonnik [~chatzilla@59.178.167.103] has quit [Quit: ChatZilla 0.9.88 [Firefox 10.0.2/20120215223356]] | 16:16 | |
-!- vikram360 [~vikram360@117.192.187.94] has quit [Ping timeout: 260 seconds] | 16:18 | |
-!- puneetgoyal [~puneet@115.242.20.205] has quit [Ping timeout: 252 seconds] | 16:18 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 16:20 | |
-!- vikram360 [~vikram360@117.192.188.25] has joined #shogun | 16:27 | |
-!- blackburn [~qdrgsm@83.234.54.68] has quit [Ping timeout: 252 seconds] | 16:28 | |
-!- gsomix [~gsomix@83.149.21.22] has quit [Ping timeout: 246 seconds] | 16:52 | |
-!- vikram360 [~vikram360@117.192.188.25] has quit [Ping timeout: 276 seconds] | 16:53 | |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 16:56 | |
-!- cwidmer [~quassel@zrcout.mskcc.org] has joined #shogun | 16:57 | |
-!- cwidmer is now known as ckwidmer | 16:57 | |
-!- ckwidmer [~quassel@zrcout.mskcc.org] has quit [Client Quit] | 16:59 | |
-!- vikram360 [~vikram360@117.192.162.252] has joined #shogun | 17:02 | |
-!- vikram360 [~vikram360@117.192.162.252] has quit [Read error: Connection reset by peer] | 17:15 | |
-!- vikram360 [~vikram360@117.192.179.147] has joined #shogun | 17:15 | |
-!- wiking [~wiking@78-23-191-201.access.telenet.be] has joined #shogun | 17:17 | |
-!- wiking [~wiking@78-23-191-201.access.telenet.be] has quit [Changing host] | 17:17 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 17:17 | |
-!- cwidmer [~quassel@zrcout.mskcc.org] has joined #shogun | 17:21 | |
-!- cwidmer [~quassel@zrcout.mskcc.org] has quit [Client Quit] | 17:21 | |
-!- ckwidmer [~quassel@mskresolve-a.mskcc.org] has joined #shogun | 17:21 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds] | 17:50 | |
-!- harshit_ [~harshit@182.68.142.173] has joined #shogun | 17:51 | |
-!- ckwidmer [~quassel@mskresolve-a.mskcc.org] has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.] | 17:56 | |
-!- mayanks43 [~Marcus@115.248.130.148] has quit [Quit: Leaving.] | 18:02 | |
-!- in3xes [~in3xes@59.163.196.121] has joined #shogun | 18:06 | |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Ping timeout: 245 seconds] | 18:17 | |
-!- in3xes [~in3xes@59.163.196.121] has quit [Ping timeout: 245 seconds] | 18:21 | |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 18:33 | |
-!- in3xes [~in3xes@59.163.196.121] has joined #shogun | 18:33 | |
-!- in3xes [~in3xes@59.163.196.121] has quit [Remote host closed the connection] | 18:37 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 18:39 | |
-!- romi_ [~mizobe@187.74.2.1] has quit [Ping timeout: 246 seconds] | 18:41 | |
-!- PhilTillet [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 18:43 | |
-!- romi_ [~mizobe@187.101.56.61] has joined #shogun | 18:54 | |
-!- harshit_ [~harshit@182.68.142.173] has quit [Ping timeout: 252 seconds] | 19:01 | |
-!- romi__ [~mizobe@187.101.63.95] has joined #shogun | 19:05 | |
-!- romi_ [~mizobe@187.101.56.61] has quit [Ping timeout: 260 seconds] | 19:05 | |
-!- harshit_ [~harshit@182.68.142.173] has joined #shogun | 19:12 | |
-!- emrecelikten [~emre@176.41.8.151] has joined #shogun | 19:16 | |
-!- gsomix [~gsomix@85.26.165.137] has joined #shogun | 19:21 | |
-!- emrecelikten [~emre@176.41.8.151] has quit [Quit: Leaving.] | 19:23 | |
-!- harshit_ [~harshit@182.68.142.173] has quit [Ping timeout: 240 seconds] | 19:28 | |
PhilTillet | Hey :) | 19:33 |
blackburn | hi | 19:46 |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Ping timeout: 245 seconds] | 19:50 | |
-!- gsomix [~gsomix@85.26.165.137] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 19:55 | |
-!- gsomix [~gsomix@85.26.165.137] has joined #shogun | 19:55 | |
-!- gsomix [~gsomix@85.26.165.137] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 20:01 | |
-!- gsomix [~gsomix@85.26.165.137] has joined #shogun | 20:01 | |
-!- harshit_ [~harshit@182.68.142.173] has joined #shogun | 20:01 | |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 20:06 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Remote host closed the connection] | 20:32 | |
-!- l0nr4n [~l0nr4n@unaffiliated/l0nr4n] has quit [Quit: This computer (or maybe me) has gone to sleep] | 20:42 | |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Ping timeout: 245 seconds] | 20:47 | |
-!- romi__ [~mizobe@187.101.63.95] has quit [Ping timeout: 246 seconds] | 20:47 | |
@sonney2k | hi | 20:54 |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has joined #shogun | 20:59 | |
-!- romi__ [~mizobe@187.101.136.131] has joined #shogun | 21:00 | |
@sonney2k | PhilTillet, any progress on opencl'ing linear combinations of kernels? | 21:03 |
PhilTillet | hello sonney2k | 21:04 |
PhilTillet | yes :) | 21:04 |
PhilTillet | I have implemented the formula with a gaussian kernel and random samples | 21:04 |
PhilTillet | but I don't really know where to start to integrate it | 21:04 |
@sonney2k | well did you benchmark it? | 21:04 |
@sonney2k | I mean that is what is most interesting for now or? | 21:05 |
PhilTillet | yes :) I wanted to benchmark it but needed a dataset | 21:05 |
shogun-buildbot | build #185 of nightly_none is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/185 | 21:05 |
@sonney2k | well random data again | 21:05 |
@sonney2k | random alphas... | 21:05 |
PhilTillet | is it normal to get k(x,xi) = 0 for random data of very high values? | 21:06 |
PhilTillet | (between 0 and 1) | 21:06 |
@sonney2k | gaussian kernel is between 0 and 1 | 21:06 |
PhilTillet | i think it is because the norm of the difference becomes infinity | 21:06 |
@sonney2k | 1 on diagonal | 21:06 |
PhilTillet | but the opencl computation is not optimized at all for now, just wanted to make it work | 21:07 |
PhilTillet | i'll benchmark it right away | 21:08 |
PhilTillet | (for now it is really really really not optimized though) | 21:09 |
PhilTillet | well | 21:14 |
PhilTillet | one second for 1000*10000 data matrix | 21:15 |
PhilTillet | but it is really not optimized | 21:15 |
shogun-buildbot | build #184 of nightly_all is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_all/builds/184 | 21:16 |
gsomix | sonney2k, hi | 21:20 |
@sonney2k | hi gsomix | 21:20 |
gsomix | it seems that python_static works | 21:22 |
@sonney2k | gsomix, nice | 21:24 |
@sonney2k | did you need to do lots of fixes for that? | 21:24 |
PhilTillet | sonney2k, do you think I should try to optimize my OpenCL code or try to somewhat integrate it ? :) | 21:24 |
@sonney2k | PhilTillet, how many alphas did you have in your setup? | 21:25 |
@sonney2k | 1000? | 21:25 |
PhilTillet | 10000 | 21:25 |
@sonney2k | and 1000 test examples of which dimensionality? | 21:25 |
PhilTillet | I think I am confused | 21:25 |
@sonney2k | you should compare this to the non opencl variant | 21:25 |
@sonney2k | when you do k(x, x_i) | 21:26 |
@sonney2k | what is the dimension of x? | 21:26 |
PhilTillet | 1000 | 21:26 |
@sonney2k | and how many examples did you test this with? 1 then? | 21:26 |
PhilTillet | Okay I think there is some point I didn't understand | 21:26 |
PhilTillet | in the formula here | 21:27 |
PhilTillet | http://shogun-toolbox.org/doc/en/current/classshogun_1_1CKernelMachine.html | 21:27 |
PhilTillet | I have : N = 10 000 and x.size() = 1 000 | 21:27 |
@sonney2k | (have to leave train soon) | 21:27 |
PhilTillet | and computed one f(x) | 21:27 |
@sonney2k | yeah that is evaluated for 1 example | 21:27 |
@sonney2k | but that is not what you want to optimize for | 21:27 |
@sonney2k | you should assume that this will be evaluated for thousands of x | 21:28 |
PhilTillet | oh okay | 21:28 |
@sonney2k | so the overhead of transfering alpha / x_i into GPU mem pays off | 21:28 |
PhilTillet | well, the transfer is at 16 GB/s , so it is more or less instant :p | 21:28 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:29 | |
PhilTillet | I think I get it | 21:29 |
PhilTillet | There is a big matrix A, and f is computed for each column of A, using the other columns as x_i | 21:29 |
PhilTillet | right? | 21:30 |
PhilTillet | (I should watch again my courses on SVM :p) | 21:30 |
@sonney2k | PhilTillet, otherwise you cannot hope to be faster at all because all there is is reading over x_i / alpha_i x once | 21:34 |
-!- l0nr4n [~l0nr4n@g225120250.adsl.alicedsl.de] has joined #shogun | 21:35 | |
-!- l0nr4n [~l0nr4n@g225120250.adsl.alicedsl.de] has quit [Changing host] | 21:35 | |
-!- l0nr4n [~l0nr4n@unaffiliated/l0nr4n] has joined #shogun | 21:35 | |
PhilTillet | ah wait, I said 1 sec but i made a huge mistake benchmarking | 21:35 |
gsomix | sonney2k, no, not much | 21:36 |
gsomix | i learned how to do git branches (tnx blackburn). | 21:39 |
harshit_ | hello sonney2k, can you please help me on minimum residual function . i found one in lapack (GMRES.f), but it is not in the lapack library in <shogun/mathematics/lapack.h> | 21:39 |
gsomix | i can try to make small commits now, instead of one large :) | 21:39 |
@sonney2k | gsomix, very good :) | 21:44 |
-!- gsomix [~gsomix@85.26.165.137] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 21:44 | |
-!- gsomix [~gsomix@85.26.165.137] has joined #shogun | 21:45 | |
-!- romi__ [~mizobe@187.101.136.131] has quit [Ping timeout: 246 seconds] | 21:46 | |
n4nd0 | sonney2k: hi there! I am sorry to insist on the same but do you think it could be possible to apply to "various usability impromevents" even if I have not experience with swig? | 21:48 |
-!- l0nr4n [~l0nr4n@unaffiliated/l0nr4n] has quit [Ping timeout: 265 seconds] | 21:48 | |
-!- l0nr4n [~l0nr4n@g225121104.adsl.alicedsl.de] has joined #shogun | 21:48 | |
-!- l0nr4n [~l0nr4n@g225121104.adsl.alicedsl.de] has quit [Changing host] | 21:48 | |
-!- l0nr4n [~l0nr4n@unaffiliated/l0nr4n] has joined #shogun | 21:48 | |
@sonney2k | n4nd0, well you should attempt to show that you can deal with swig then | 21:48 |
n4nd0 | sonney2k: my first option is SO but I think it might be better if I apply to another project as well, just to get better odds | 21:48 |
@sonney2k | n4nd0, do you know of anyone else applying for SO? | 21:49 |
n4nd0 | sonney2k: I have not heard of anyone, neither here nor in the mailing list | 21:49 |
@sonney2k | n4nd0, regarding how to show - write some small typemap - e.g. the one missing in octave modular... | 21:50 |
@sonney2k | ...have to leave train. | 21:50 |
n4nd0 | sonney2k: ok, see you later | 21:50 |
-!- blackburn [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Ping timeout: 245 seconds] | 21:52 | |
n4nd0 | harshit_: hey! | 21:59 |
n4nd0 | harshit_: I read your question, have you in any case tried to use it? | 21:59 |
harshit_ | hey n4nd0 | 22:00 |
n4nd0 | harshit_: I think it is not necessary for it to be defined in shogun/mathematics/lapack.h, you may use the function from the lapack library | 22:00 |
harshit_ | but lapack library is in fortran, how can i use it in c++ | 22:00 |
n4nd0 | harshit_: one thing is the language the code is written and another the library | 22:01 |
n4nd0 | harshit_: our C++ code in shogun is used from python, java, matlab, octave etc using swig for example | 22:01 |
harshit_ | okay, havnt done that before | 22:02 |
n4nd0 | harshit_: if you look at lapack.cpp | 22:02 |
n4nd0 | harshit_: that's C++ code, but is it calling the functions from lapack | 22:02 |
harshit_ | okay | 22:03 |
harshit_ | so in that sense i can add new | 22:03 |
harshit_ | functions in lapack.cpp | 22:03 |
n4nd0 | in any case, I am not completely sure about this | 22:03 |
harshit_ | and then use it in my program | 22:03 |
n4nd0 | I recommend you to try | 22:03 |
n4nd0 | yes, you could do that | 22:03 |
harshit_ | i'll try that | 22:04 |
harshit_ | btw which project are you going for in gsoc | 22:05 |
n4nd0 | SO | 22:05 |
n4nd0 | structured output | 22:05 |
-!- akarshanSK [~akarshan@202.3.77.217] has joined #shogun | 22:06 | |
n4nd0 | you? | 22:06 |
harshit_ | newtonSVM and integration of new versions of liblinear | 22:07 |
harshit_ | also i was thinking to give another proposal for C5.0 | 22:07 |
harshit_ | decision trees | 22:08 |
n4nd0 | ok | 22:08 |
harshit_ | is that possible, To give 2 proposals to same organization | 22:08 |
n4nd0 | the deep learning stuff didn't succeed finally? | 22:08 |
n4nd0 | yes, no problem with that | 22:08 |
n4nd0 | the only limitation is 20 proposals | 22:08 |
n4nd0 | you can distribute them as you wish | 22:09 |
harshit_ | nah, didn't get any update on that | 22:09 |
harshit_ | also there are not much libraries for deep learning | 22:10 |
harshit_ | that can be integrated into shogun | 22:10 |
harshit_ | Any other organization you applying for ? | 22:11 |
n4nd0 | no | 22:15 |
n4nd0 | you? | 22:15 |
-!- harshit_ [~harshit@182.68.142.173] has quit [Ping timeout: 260 seconds] | 22:17 | |
-!- harshit_ [~harshit@182.68.142.173] has joined #shogun | 22:22 | |
harshit_ | @n4nd0 no, thought of applying to orange, but then dropped the idea | 22:24 |
-!- harshit_ [~harshit@182.68.142.173] has quit [Quit: Leaving] | 22:33 | |
-!- puffin444 [230bf329@gateway/web/freenode/ip.35.11.243.41] has joined #shogun | 22:37 | |
-!- blackburn [~qdrgsm@83.234.54.68] has joined #shogun | 22:38 | |
gsomix | sonney2k, hey? | 22:40 |
-!- akarshanSK [~akarshan@202.3.77.217] has quit [Quit: Leaving] | 22:40 | |
@sonney2k | gsomix, yes? | 22:42 |
blackburn | sonney2k: I need to discuss my issue ;) | 22:43 |
blackburn | n4nd0: hey I'm here finally | 22:43 |
blackburn | n4nd0: why do you want to abandon SO idea? | 22:44 |
gsomix | sonney2k, i have problem in "./configure" code. | 22:44 |
-!- puffin444 [230bf329@gateway/web/freenode/ip.35.11.243.41] has left #shogun [] | 22:45 | |
-!- Miggy [~piggy@14.139.82.6] has joined #shogun | 22:45 | |
gsomix | this is related to get flags for linking python's library | 22:46 |
@sonney2k | gsomix, can you be more specific? | 22:46 |
-!- romi_ [~mizobe@187.57.1.50] has joined #shogun | 22:47 | |
gsomix | sonney2k, distutils.sysconfig.get_config_vars('SO') for python3 gives a result different from python2 | 22:48 |
Miggy | sonney2k:i am interested in working on the project of implementing decision trees...so how should i proceed with it? | 22:48 |
gsomix | sonney2k, see ". / configure" on line 4302. the results are "?python-32mu.so" for python3 and ".so" for python2 | 22:50 |
CIA-64 | shogun: Sergey Lisitsyn master * r2002a08 / src/interfaces/modular/Library.i : Late fix for java serialization dummies - http://git.io/KUjQSw | 22:50 |
blackburn | shogun-buildbot: work! | 22:51 |
shogun-buildbot | What you say! | 22:51 |
PhilTillet | lol | 22:51 |
Miggy | blackburn: which project are u choosing this time? :P | 22:51 |
@sonney2k | gsomix, maybe you can workaround this by doing distutils.sysconfig.get_config_vars('SO')[0].rsplit('.')[-1] | 22:51 |
blackburn | Miggy: I want to work on domain adaptation and MT learning | 22:51 |
Miggy | cool | 22:52 |
blackburn | Miggy: you are asking in a way that makes me think you know who I am and what is I was working on before ;) | 22:52 |
Miggy | yes i know u worked on dimensionality reduction before :) | 22:53 |
gsomix | sonney2k, may be necessary to use distutils.sysconfig.get_config_vars('BLDLIBRARY')? it returns "-L. -lpython3.2" | 22:54 |
@sonney2k | gsomix, -L. is not what we want though | 22:54 |
Miggy | sonney2k: waiting for ur reply | 22:55 |
blackburn | Miggy: while Soeren is in ignore mode I would suggest you to contribute with something small | 22:55 |
blackburn | :D | 22:55 |
Miggy | hehe :) | 22:55 |
Miggy | related to decision tress or in general? | 22:55 |
blackburn | it is mainly up to you | 22:56 |
blackburn | decision trees in C++ can take a while | 22:56 |
Miggy | yes | 22:57 |
@sonney2k | gsomix, look http://projects.scipy.org/numpy/ticket/1749 - scipy people had the same problem | 22:57 |
blackburn | Miggy: were you applying last year? | 22:57 |
Miggy | yes | 22:57 |
blackburn | aha I see | 22:58 |
Miggy | only 5 slots :( | 22:58 |
blackburn | yeah it was tight | 22:58 |
PhilTillet | sonney2k, I made a mistake on my benchmark, the result was 1ms and not 1second :p | 22:58 |
Miggy | this time expecting more? | 22:58 |
blackburn | this year we expect a little more | 22:58 |
Miggy | great | 22:58 |
blackburn | no reason to really expect more although :D but makes sense cause we finished with glance | 22:59 |
gsomix | sonney2k, tnx | 22:59 |
blackburn | shogun-buildbot: !help | 22:59 |
blackburn | :D | 23:00 |
Miggy | the main problem is with the priority of the projects...some project u like to work on does not get selected :( | 23:00 |
blackburn | Miggy: I hope most interesting projects would be covered this year | 23:01 |
@sonney2k | PhilTillet, what would be really interesting is to see a comparison to CPU version and for multiple x ... | 23:02 |
blackburn | what can I say is some file formats idea/mloss idea/ update ocas have lower priority | 23:02 |
blackburn | actually ocas is up-to-date here | 23:02 |
blackburn | :D | 23:02 |
PhilTillet | sonney2k, yes, sure, but I think the computation of a single x with the parallel algorithm is way faster | 23:02 |
@sonney2k | I don't think so | 23:03 |
@sonney2k | why should it? | 23:03 |
PhilTillet | because everything is done in a parallel maner | 23:03 |
Miggy | blackburn: ok so i want to know one thing...the previous 5 students can also apply this time...so how can a new student get chance? | 23:03 |
PhilTillet | with a looot of threads | 23:03 |
blackburn | Miggy: only 2 of us are applying | 23:03 |
@sonney2k | PhilTillet, yes but bottleneck should be memory accesses | 23:04 |
Miggy | u and? | 23:04 |
@sonney2k | so it doesn't matter if GPU or CPU is waiting | 23:04 |
blackburn | Heiko | 23:04 |
Miggy | rest? | 23:04 |
blackburn | Miggy: actually we have equal chances :) | 23:04 |
blackburn | I mean each should proceed in standart way | 23:04 |
PhilTillet | sonney2k, Oh I understand your point | 23:04 |
blackburn | Miggy: rest have dissappeared :D | 23:04 |
blackburn | s* | 23:05 |
@sonney2k | PhilTillet, that is why I am saying try for multiple x (say 10k or more) | 23:05 |
Miggy | i mean they will definitely pick you because u have worked once with them | 23:05 |
@sonney2k | and dim(x) = 100 or so | 23:05 |
blackburn | Miggy: even in this case we would have 6-7 slots and there is a room | 23:05 |
-!- gsomix [~gsomix@85.26.165.137] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 23:06 | |
PhilTillet | sonney2k, are you talking about the kernel function? | 23:06 |
Miggy | i see | 23:06 |
PhilTillet | http://shogun-toolbox.org/doc/en/current/classshogun_1_1CKernelMachine.html#adef2f22ca9c677e60635bbab682e117e . I thought on this formula the number of examples was N | 23:07 |
PhilTillet | this is why I am confused | 23:07 |
Miggy | i hope its not too late to contribute | 23:07 |
@sonney2k | PhilTillet, the number of support vectors is N | 23:07 |
@sonney2k | the number of vectors this formula is applied to is just 1 | 23:07 |
@sonney2k | (just one x) | 23:07 |
PhilTillet | Oh, I get it :p | 23:07 |
-!- vikram360 [~vikram360@117.192.179.147] has quit [Ping timeout: 260 seconds] | 23:07 | |
blackburn | Miggy: sure, not late | 23:08 |
Miggy | so how many students are currently active on shogun? | 23:09 |
Miggy | this would give me an idea of my standing :P | 23:09 |
blackburn | Miggy: we've got contributions from 3 students already | 23:10 |
blackburn | (w/o me and Heiko) | 23:10 |
blackburn | last year there was a simple task which I come up with | 23:10 |
blackburn | kernel functions | 23:10 |
blackburn | :) | 23:10 |
PhilTillet | I understand sonney2k, it's like the difference between a matrix-vector and a matrix-matrix product :) | 23:11 |
Miggy | ok so can u suggest me something? | 23:11 |
Miggy | what type of contributions have they done? | 23:11 |
blackburn | Miggy: ok let me try | 23:11 |
blackburn | wiking has done HomogeneousKernelMap preprocessor | 23:11 |
blackburn | n4nd0 was working on qda | 23:12 |
blackburn | gsomix are working on python3 | 23:12 |
blackburn | Miggy: how familiar you are with shogun's internals? | 23:12 |
Miggy | had a look last year | 23:13 |
Miggy | but was able to code comfortably | 23:13 |
blackburn | I am pretty interested with least angle regression but no idea if it would be appropriate to suggest | 23:14 |
Miggy | smthing in classification or clustering? | 23:15 |
blackburn | Miggy: some dataset downloader doing classification/regression and printing out results would be nice | 23:15 |
PhilTillet | sonney2k, support vectors and examples are stored columnwise, right ? | 23:16 |
Miggy | ok will look into that | 23:16 |
@sonney2k | PhilTillet, yes | 23:16 |
PhilTillet | okay :) | 23:17 |
blackburn | Miggy: some simple extension like new kernel/distance are welcome as well | 23:17 |
Miggy | all the kernel functions have been already done :P | 23:17 |
blackburn | yeah even mahalanobis (which I forgot) | 23:18 |
blackburn | was done by n4nd0 | 23:18 |
Miggy | so prvious year students who g | 23:18 |
Miggy | sry | 23:18 |
blackburn | ;) | 23:18 |
Miggy | previous year students also applying this time? | 23:18 |
Miggy | the one who contributed previously? | 23:19 |
blackburn | no, only 2 | 23:19 |
blackburn | we both have been contributing sinch last pre-gsoc period | 23:19 |
Miggy | i mean those who did smthing but were not selected | 23:19 |
blackburn | I think you are the first one of that kind | 23:19 |
Miggy | well i was late so i was not abe to give any patch that time | 23:20 |
blackburn | I see | 23:20 |
Miggy | i got to know abt shogun a little late :/ | 23:20 |
blackburn | sonney2k: sudo answer me | 23:20 |
blackburn | damn is he kidding | 23:21 |
wiking | blackburn: don't forget the great jensen kernel :))) | 23:21 |
blackburn | oh yeah, Miggy: wiking has also done Jensen-Shannon kernel | 23:21 |
wiking | :P | 23:22 |
blackburn | I have not been aware of it for whole life | 23:22 |
blackburn | ;) | 23:22 |
Miggy | err need some good patch | 23:23 |
Miggy | http://crsouza.blogspot.in/2010/03/kernel-functions-for-machine-learning.html | 23:23 |
blackburn | wiking: in fact I did not manage to use J-S explicitly | 23:23 |
Miggy | all have been done :P | 23:23 |
blackburn | it takes %INFINITY% time | 23:23 |
wiking | hahahaha | 23:23 |
blackburn | but marvellous homogay :D map is awesome | 23:24 |
wiking | but then i guess with the homomap was good ;) | 23:24 |
wiking | :P | 23:24 |
blackburn | best results are with intersection | 23:24 |
blackburn | no idea why | 23:24 |
wiking | mmm i've seen somebody today sending an email wanting to do the latent thingy | 23:24 |
blackburn | wiking: well you are a strong candidate on latent svms :D | 23:25 |
wiking | hehehhe :> | 23:25 |
wiking | i have to do some work with it | 23:25 |
wiking | but now i have so much stupid distractions | 23:25 |
wiking | i might be able to do it tomorrow | 23:25 |
n4nd0 | blackburn: not abandoning SO man, no way :) | 23:25 |
wiking | at least class definitions | 23:25 |
n4nd0 | blackburn: I am really investing time on there | 23:25 |
blackburn | n4nd0: so what is the problem? | 23:26 |
n4nd0 | blackburn: none, I thought it could be better to apply to a second project just in case SO doesn't get selected finally or another studen takes it | 23:26 |
n4nd0 | blackburn: what do you think? | 23:26 |
blackburn | n4nd0: ah sure | 23:26 |
n4nd0 | blackburn: I want to do SO for sure, but I'd like to get better chances | 23:26 |
blackburn | n4nd0: then you may also apply for ECOC/MC idea | 23:27 |
n4nd0 | blackburn: let me check | 23:27 |
blackburn | wiking: btw in this vedaldi paper there are map for gaussian kernel, right? | 23:28 |
wiking | blackburn: not that i know of | 23:28 |
wiking | or i don't remember it having | 23:28 |
n4nd0 | blackburn: which one is that? ECOC/MC | 23:28 |
blackburn | are there maps only for additive kernels? | 23:28 |
blackburn | n4nd0: multiclass | 23:28 |
blackburn | mentored by Soeren | 23:29 |
n4nd0 | blackburn: isn't it that your target? | 23:29 |
blackburn | n4nd0: no | 23:29 |
n4nd0 | blackburn: aham ... I thought that! | 23:29 |
-!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 23:29 | |
blackburn | n4nd0: noo, I apply for mt/da | 23:29 |
n4nd0 | blackburn: I have not even read that because of that :P | 23:29 |
blackburn | :D | 23:30 |
blackburn | I apply for multitask learning with christian | 23:30 |
-!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has quit [Client Quit] | 23:30 | |
n4nd0 | blackburn: all right, I will check that other then before starting to see how swig works | 23:30 |
n4nd0 | blackburn: anyhow, time to fix the repo ... | 23:31 |
blackburn | oh damn yeah | 23:31 |
n4nd0 | blackburn: so do you think the best is to kill the fork and do it again? | 23:31 |
blackburn | n4nd0: yeah the easiest way for me | 23:31 |
blackburn | n4nd0: swig would be competetive cause there is a student wants to work on that already | 23:32 |
blackburn | but multitask is pretty free :) | 23:32 |
blackburn | err | 23:32 |
blackburn | multiclass | 23:32 |
blackburn | multimulti | 23:32 |
blackburn | :D | 23:32 |
blackburn | multipulti | 23:32 |
n4nd0 | blackburn: ok good tip man, thank you very much :) | 23:32 |
n4nd0 | haha | 23:32 |
n4nd0 | blackburn: you ok :P? | 23:32 |
blackburn | yeah | 23:33 |
blackburn | ok late here -> sleep time | 23:34 |
n4nd0 | :) | 23:34 |
n4nd0 | good night then | 23:34 |
blackburn | good night | 23:34 |
blackburn | sonney2k: sudo ignore-mode-off! | 23:34 |
blackburn | n4nd0: kick him when he comes | 23:35 |
blackburn | see you | 23:35 |
n4nd0 | blackburn: we'll see | 23:35 |
n4nd0 | bye | 23:35 |
-!- blackburn [~qdrgsm@83.234.54.68] has quit [Ping timeout: 276 seconds] | 23:40 | |
shogun-buildbot | build #384 of java_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/java_modular/builds/384 | 23:44 |
PhilTillet | hello, shogun-buildbot ! tired to be a bot? :/ | 23:49 |
n4nd0 | haha | 23:50 |
n4nd0 | shogun-buildbot: work! | 23:50 |
shogun-buildbot | What you say! | 23:50 |
PhilTillet | o_o | 23:50 |
n4nd0 | he has worked enought today maybe ... | 23:50 |
n4nd0 | he/she | 23:50 |
n4nd0 | :) | 23:51 |
PhilTillet | yes :) | 23:51 |
--- Log closed Thu Mar 22 00:00:19 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!