IRC logs of #shogun for Thursday, 2011-12-08

« Wednesday, 2011-12-07 Index Friday, 2011-12-09 »
--- Log opened Thu Dec 08 00:00:19 2011
-!- blackburn [~blackburn@188.168.2.160] has quit [Quit: Leaving.]01:31
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun02:44
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]02:47
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]04:16
-!- Ram108 [~amma@14.99.41.10] has joined #shogun07:06
Ram108hi07:06
Ram108i am new to shogun and i would like to contribute something.....07:08
Ram108could someone guide me as to where to start...... i have tried google-ing up a lot of key words but could not get much idea out of it07:09
Ram108thank you....07:10
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun08:44
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed]09:42
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun10:06
-!- Ram108 [~amma@14.99.41.10] has quit [Ping timeout: 252 seconds]11:25
-!- kocki [8d2c1705@gateway/web/freenode/ip.141.44.23.5] has joined #shogun12:04
kockihello12:28
kockii have a question12:29
kockimay i link to shogun from my application without releasing it under the gpl?12:30
sonne|workwell you can use any gpl compatible license for your code13:05
sonne|workkocki: ^13:05
sonne|workbut you have to opensource your code  under a compat license if you want to distribute the conglomerate product13:07
kockiok, thank you13:07
kockii'm afraid, that won't be possible13:08
kockithen, i'll have to use libsvm or some other library13:09
sonne|workkocki: look at scikits learn or torch or browse mloss.org13:13
sonne|workmaybe you'll find sth13:13
kockithank you very much! i'll do that13:14
kockiscikit learn is for python, but i need something for c++13:16
-!- blackburn [~blackburn@188.168.14.102] has joined #shogun13:17
sonne|workkocki: dlib ml?13:25
sonne|workanyways no idea really what is best for you13:26
kockii'm looking at dlib at the moment. looks promising13:28
blackburnsonne|work: btw I don't really understand why we are GPL ;)13:33
sonne|workhehe13:33
blackburnsonne|work: testing now13:34
blackburnRam108: if you would read logs you should contact me with blackburn91@gmail.com13:35
15SAAI18Mshogun: Sergey Lisitsyn master * r52f7c5a / data : Update data - http://git.io/dB580Q13:43
blackburnsonne|work: did you already turned off trac?13:44
sonne|worknot intentionally13:44
sonne|workis it off?13:44
sonne|worklooks like13:45
sonne|workhmhh13:45
sonne|workwell then one more reason to migrate :D13:45
sonne|workI have to check why it is not running but cannot do that from here13:45
blackburnsonne|work: I guess your kids did13:45
blackburnsonne|work: I hardly remember all the issues ;)13:45
sonne|workwell then make a new plan for 1.2 - right?13:46
sonne|worklike swig 2.0.4 as hard requirement and stuff13:46
shogun-buildbotbuild #388 of cmdline_static is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/38813:47
blackburnsonney2k: I'm not sure it is an issue13:47
shogun-buildbotbuild #363 of python_static is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/36313:48
blackburnsonne|work: http://dl.dropbox.com/u/10139213/shogun/words.png did I show it to you?13:53
blackburnLLE with 'diff' Custom kernel defined on words13:53
blackburnKernel LLE in fact13:54
15SAAI18Mshogun: Sergey Lisitsyn master * rf77788a / applications/dimreduction/words_embedding.py : Updated words embedding application - http://git.io/cTpkLw14:00
kockii just want to thank you and tell you that i will try dlib. bye14:21
sonne|workkocki: ok - np14:28
sonne|workblackburn: which kernel is this?14:28
sonne|workCommWordStringKernel with sorted preproc?14:28
blackburnsonne|work: custom14:48
blackburnfor i in range(N):14:48
blackburnfor j in range(i,N):14:48
blackburns = difflib.SequenceMatcher(None,words[i],words[j])14:48
blackburn:)14:48
blackburndist_matrix[i,j] = s.ratio()14:48
blackburnshould be kernel_matrix ;)14:49
sonne|workI would suggest to just use CommWordStringKernel for that ... no idea what sequence matcher does though14:53
sonne|workthe Comm* kernel will count e.g. 2 mers14:53
blackburnsonne|work: I'll try14:53
-!- kocki [8d2c1705@gateway/web/freenode/ip.141.44.23.5] has quit [Quit: Page closed]14:54
blackburnsonne|work: how to use it? :)14:59
blackburngot it15:11
blackburnbut results are worse or I set wrong params15:13
sonne|workI need to see the code15:21
blackburncf = StringCharFeatures(words, RAWBYTE)15:23
blackburnwf = StringWordFeatures(RAWBYTE)15:23
blackburnwf.obtain_from_char(cf,0,1,2,False)15:23
blackburnreturn CommWordStringKernel(wf,wf)15:23
blackburnsonne|work: words are ['publishing','below',...]15:23
blackburnyou said something about sorted preprocs?15:23
sonne|workblackburn: yes look in the example kernel_commwo*15:24
sonne|workotherwise result will be crap15:25
blackburnyeah it is now15:25
blackburnsonne|work: what is order,start and gap?15:25
sonne|workorder is ngram length (use 2)15:26
sonne|workstart is usually order-115:26
sonne|workand gab is not used15:26
sonne|workso IIRC 015:26
blackburnsonne|work: hmm pretty interesting results too15:27
sonne|workyou could use CommUlongStringKernel instead15:27
sonne|workand the up to order 415:28
blackburnsonne|work: got something like that http://dl.dropbox.com/u/10139213/shogun/words.png15:34
sonne|workworks well right?15:44
blackburnsonne|work: sure15:48
sonne|workwas that order 4?15:49
sonne|workor 2?15:49
blackburnsonne|work: 2, 4 wasn't working at all15:49
sonne|work4 is too tough I guess15:49
sonne|workthere are rarely 3-mer matches15:49
sonne|workwe should have hashed string features too15:50
sonne|workbut currently these only exist for DNA15:50
-!- blackburn [~blackburn@188.168.14.102] has left #shogun []19:50
-!- blackburn [~blackburn@188.168.14.102] has joined #shogun19:57
blackburnsonney2k: turn on trac please ;)20:54
-!- song [9810375e@gateway/web/freenode/ip.152.16.55.94] has joined #shogun23:21
songHi  I was trying to use R-shogun interface with the example script: kernel_oligo_string_modular.R. I got the following error when excuting line 16:  > dump <- feats_train$set_features(feats_train, fm_train_dna) Error in feats_train$set_features(feats_train, fm_train_dna) :   could not find function "f"  Could anyone help me to interpret this error message?23:21
-!- song is now known as Guest3242623:21
Guest32426Hi  I was trying to use R-shogun interface with the example script: kernel_oligo_string_modular.R. I got the following error when excuting line 16:  > dump <- feats_train$set_features(feats_train, fm_train_dna) Error in feats_train$set_features(feats_train, fm_train_dna) :   could not find function "f"  Could anyone help me to interpret this error message?23:21
Guest32426i cannot send question to the shogun maillist so I post it here.23:22
-!- Guest32426 [9810375e@gateway/web/freenode/ip.152.16.55.94] has quit [Client Quit]23:22
--- Log closed Fri Dec 09 00:00:19 2011
« Wednesday, 2011-12-07 Index Friday, 2011-12-09 »

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!