--- Log opened Wed Apr 18 00:00:19 2012 | ||
blackburn | sonney2k: no idea.. | 00:00 |
---|---|---|
blackburn | there should be a reason though | 00:00 |
@sonney2k | hmmhh maybe because of LIbSVMMultiClass | 00:00 |
blackburn | SGVector<int32_t> &svm_svs() 142+ { 143+ return svm_proto()->m_svs; 144+ } | 00:03 |
blackburn | sonney2k: may be that ^? | 00:03 |
@sonney2k | m_machines = SGVector<CMachine *>(num_svms); | 00:03 |
@sonney2k | that is sth I don't like | 00:03 |
@sonney2k | all the rest looks pretty nice to me | 00:03 |
blackburn | sonney2k: why? | 00:03 |
blackburn | I did it in the same way | 00:03 |
blackburn | sonney2k: m_machines = create_machines(num_svms)? | 00:04 |
@sonney2k | blackburn, SGVector is a numeric thing | 00:05 |
@sonney2k | e.g. SGVector a,b | 00:05 |
@sonney2k | a+b | 00:05 |
@sonney2k | will work at some point | 00:05 |
blackburn | hmm yes | 00:05 |
blackburn | but I prefer even stl vector here | 00:05 |
@sonney2k | any of the Dyn*Arrays is what you should use here | 00:05 |
-!- emrecelikten [~Anubis@176.40.226.62] has quit [Ping timeout: 276 seconds] | 00:06 | |
blackburn | sonney2k: have you heard of valarray btw? | 00:06 |
blackburn | that's the thing we wouldn't be able to ignore in the future | 00:06 |
@sonney2k | not sure if this comes too late for us | 00:10 |
blackburn | sonney2k: we can always change underlying sgvector stuff | 00:11 |
blackburn | sonney2k: I think it should be as fast if we provide a facade like sgvector | 00:13 |
@sonney2k | true | 00:13 |
@sonney2k | so today will be the anouncement of duplicate students | 00:13 |
@sonney2k | blackburn, I gave some comments to the multiclasssvm stuff | 00:13 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 00:14 | |
@sonney2k | problem is really that this is a huge very intrusive patch | 00:14 |
blackburn | sonney2k: yes I see mails | 00:14 |
@sonney2k | very hard to look at | 00:14 |
@sonney2k | but I think pluskid did a nice job | 00:14 |
blackburn | true | 00:14 |
@sonney2k | maybe some of the multiclass svm proxy stuff can go away but apart from the minor changes (pending that it all still passes test!) it has my OK | 00:14 |
blackburn | sonney2k: I would like to merge it soon | 00:15 |
@sonney2k | IIRC he is away now so would be good if you can give him feedback *before* he returns | 00:15 |
@sonney2k | such that he has a fresh mind and does the final touches and we can quickly merge afterwards | 00:15 |
blackburn | sonney2k: ok sure | 00:16 |
@sonney2k | blackburn, btw isn't it 4am or so at your place? | 00:16 |
@sonney2k | sun rising already? | 00:16 |
blackburn | sonney2k: hmm 2.16 | 00:16 |
@sonney2k | ahh not soo late then | 00:16 |
@sonney2k | ok | 00:16 |
@sonney2k | :D | 00:16 |
blackburn | sonney2k: I am just +2 from you I guess? | 00:17 |
blackburn | not so far away actually | 00:17 |
blackburn | in means of russian distances lol | 00:17 |
blackburn | if I was in omsk it would be +4 | 00:18 |
@sonney2k | blackburn, btw I tried to do some meeting planning - but it is difficult with so many timezones | 00:18 |
@sonney2k | someone will have to stay awake late or wake up early | 00:18 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 00:19 | |
blackburn | sonney2k: after 23? | 00:19 |
@sonney2k | problem is we have mentos in NY / melboune | 00:19 |
@sonney2k | yeah | 00:19 |
@sonney2k | all the rest is inbetween | 00:19 |
blackburn | sonney2k: actually I am not sure chris in NY right now | 00:19 |
blackburn | is in * | 00:19 |
blackburn | sonney2k: he visited tuebingen for some deals so I don't know where he is right now | 00:20 |
blackburn | but melbourne is a problem yeah | 00:20 |
blackburn | (just like cheng soon for me) :D | 00:20 |
@sonney2k | so only viable time is UTC 11 or 12hrs | 00:20 |
@sonney2k | ok | 00:21 |
@sonney2k | blackburn, how can we get rid of the 10 open pull requests? | 00:21 |
blackburn | sonney2k: we don't have to | 00:21 |
blackburn | just live with it :D | 00:21 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Client Quit] | 00:22 | |
@sonney2k | blackburn, you don't loose overview? | 00:22 |
blackburn | sonney2k: yes that can be hard but I believe it should be as it is | 00:23 |
@sonney2k | blackburn, for exampel your lib/external stuff could easily be merged if you kept Tron at its place | 00:23 |
@sonney2k | blackburn, I would rather close things we don't merge within a few days / where there is no obvious progress for some time | 00:24 |
@sonney2k | it doesn't help to have such requests open | 00:24 |
blackburn | sonney2k: I agree - my issues are slow because of my exam week | 00:24 |
-!- cronor_ [~cronor@g231230224.adsl.alicedsl.de] has joined #shogun | 00:26 | |
-!- cronor [~cronor@e178176008.adsl.alicedsl.de] has quit [Ping timeout: 276 seconds] | 00:27 | |
-!- cronor_ is now known as cronor | 00:27 | |
CIA-64 | shogun: Victor Sadkov master * r9a4214c / (7 files in 5 dirs): Merge branch 'master' of git://github.com/shogun-toolbox/shogun into refactoring - http://git.io/L8TSNA | 00:27 |
CIA-64 | shogun: Victor Sadkov master * rb886236 / src/README.developer : Fixed outdated file. - http://git.io/3U6RXA | 00:27 |
CIA-64 | shogun: Victor Sadkov master * r5f73ad6 / (2 files): Added new sample code - http://git.io/qseh-g | 00:27 |
CIA-64 | shogun: Soeren Sonnenburg master * r024fc0a / (3 files in 2 dirs): | 00:27 |
CIA-64 | shogun: Merge pull request #463 from Victor-Sadkov/master | 00:27 |
CIA-64 | shogun: New statistics sample and various whitespace changes. - http://git.io/OXSWTA | 00:27 |
@sonney2k | blackburn, this week or next - when do you have them? | 00:27 |
blackburn | sonney2k: *pre-exam sorry | 00:28 |
blackburn | I have one on 20, and on 27, 4 and 8 | 00:28 |
blackburn | sonney2k: however this week is much more hard for me than next will be | 00:29 |
blackburn | I have to get done all the things I haven't done before :D | 00:29 |
@sonney2k | great - I guess the kickoff meeting then will have only 70% of the participants | 00:30 |
@sonney2k | well we have chat logs | 00:30 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 00:31 | |
@sonney2k | alright | 00:31 |
@sonney2k | I think I should consider that sleep is an option | 00:31 |
@sonney2k | blackburn, nite | 00:31 |
blackburn | sonney2k: best option | 00:31 |
blackburn | good night | 00:31 |
blackburn | but I have a couple of things to do still heh | 00:32 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 00:44 | |
-!- harshit_ [~harshit@182.68.246.67] has quit [Quit: Leaving] | 00:56 | |
-!- av3ngr [av3ngr@nat/redhat/x-hcgxbjfacbxlnpuz] has joined #shogun | 01:47 | |
shogun-buildbot | build #489 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/489 | 04:47 |
-!- cronor [~cronor@g231230224.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 05:24 | |
shogun-buildbot | build #490 of python_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/490 blamelist: sonne@debian.org, sadkov993@gmail.com | 05:39 |
-!- cronor [~cronor@g231230224.adsl.alicedsl.de] has joined #shogun | 05:43 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 07:51 | |
n4nd0 | good morning! | 07:52 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Remote host closed the connection] | 07:55 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 08:01 | |
-!- blackburn [~qdrgsm@188.168.2.179] has quit [Quit: Leaving.] | 08:23 | |
sonne|work | n4nd0: nice one too! | 08:35 |
-!- uricamic [9320543b@gateway/web/freenode/ip.147.32.84.59] has joined #shogun | 08:46 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 08:46 | |
sonne|work | moin blackburn! | 08:49 |
blackburn | sonne|work: morgen | 08:52 |
-!- wiking [~wiking@78-23-189-112.access.telenet.be] has joined #shogun | 08:54 | |
-!- wiking [~wiking@78-23-189-112.access.telenet.be] has quit [Changing host] | 08:54 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 08:54 | |
-!- Marty28 [~Marty@158.181.78.199] has joined #shogun | 08:56 | |
Marty28 | hiho | 08:58 |
n4nd0 | hey Marty28 | 08:59 |
Marty28 | Does poim.py have any chance of running, now? | 09:02 |
n4nd0 | Marty28: it looks like sonne|work fixed it yesterday | 09:04 |
Marty28 | nice | 09:04 |
n4nd0 | as he said, it should be ready | 09:04 |
Marty28 | Do you know if shogun has a simple random data generator included? | 09:05 |
Marty28 | A data converter for different file formats would also be nice. | 09:05 |
Marty28 | I am currently writing a simple generator for DNA/RNA. | 09:06 |
n4nd0 | Marty28: take a look at random methods in CMath and SGVector | 09:06 |
Marty28 | Some code is in the examples as well. | 09:08 |
blackburn | sonne|work: I believe it is unfair that they announce on 23th of April | 09:27 |
sonne|work | ? | 09:27 |
blackburn | that should be friday (not enough room for vodking) | 09:27 |
blackburn | ;) | 09:27 |
sonne|work | hehe | 09:27 |
sonne|work | Marty28: mldata.org has mldata-utils converting between various formats | 09:27 |
Marty28 | sonne: thx | 09:29 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: Lost terminal] | 09:32 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:34 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 09:53 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 09:56 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:59 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 10:02 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Remote host closed the connection] | 10:02 | |
-!- wiking [~wiking@vpnc078.ugent.be] has joined #shogun | 10:02 | |
-!- wiking [~wiking@vpnc078.ugent.be] has quit [Changing host] | 10:02 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 10:02 | |
-!- av3ngr [av3ngr@nat/redhat/x-hcgxbjfacbxlnpuz] has quit [Quit: That's all folks!] | 10:19 | |
-!- Marina [55b326f1@gateway/web/freenode/ip.85.179.38.241] has joined #shogun | 10:43 | |
-!- Marina [55b326f1@gateway/web/freenode/ip.85.179.38.241] has quit [Client Quit] | 10:44 | |
-!- Marina [55b326f1@gateway/web/freenode/ip.85.179.38.241] has joined #shogun | 10:44 | |
sonne|work | Marina: try again :D | 10:47 |
-!- harshit_ [~harshit@182.68.167.125] has joined #shogun | 10:48 | |
wiking | sonney2k blackburn any of you here? is there anything more required for defining a new feature type than setting the EFeatureType and EFeatureClass ? | 10:52 |
blackburn | wiking: nothing more | 10:53 |
wiking | i wonder why do i get this error then base/class_list.cpp:427:93: error: allocating an object of abstract class type 'CLatentFeatures' | 10:54 |
wiking | afaik i've implemented all the =0 functions... | 10:54 |
blackburn | wiking: get_name? | 10:54 |
wiking | yep i have that as well | 10:54 |
blackburn | wiking: do you inherit it from dotfeatures? | 10:54 |
wiking | i have duplicate, get_num_vectors, get_size, get_name, get_feature_type and get_feature_class implemented | 10:55 |
wiking | now i've inherited from cfeatures, just for simplicity... | 10:55 |
blackburn | hmhm | 10:55 |
wiking | and i still get this | 10:56 |
wiking | base/class_list.cpp:427:93: error: allocating an object of abstract class type 'CLatentFeatures' | 10:56 |
wiking | static CSGObject* __new_CLatentFeatures(EPrimitiveType g) { return g == PT_NOT_GENERIC? new CLatentFeatures(): NULL; } | 10:56 |
wiking | i've tried a make clean && make | 10:56 |
wiking | but did not help | 10:56 |
wiking | i really wonder now what other pure virtual functions are there that i haven't implemented yet in cfeatures | 10:57 |
wiking | i guess i'll check it now in the source | 10:57 |
blackburn | the only virtual of SGObject is get_name | 10:58 |
blackburn | so you have to implement only get_name + virtual funcs of cfeatures | 10:58 |
wiking | yep | 10:58 |
wiking | afaik i did it | 10:58 |
wiking | :) | 10:58 |
blackburn | wiking: https://github.com/shogun-toolbox/shogun/blob/master/src/shogun/features/DummyFeatures.cpp | 10:59 |
wiking | yep | 11:02 |
wiking | have them :) | 11:02 |
wiking | http://snipt.org/uhhn1#expand | 11:02 |
wiking | my definition | 11:02 |
harshit_ | hello everyone :) | 11:07 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 11:12 | |
wiking | blackburn: any ideas? | 11:17 |
wiking | i really couldn't find more pure virtual functions that i'm missing and no other things i would miss... but for sure it's something with my implementation | 11:18 |
blackburn | wiking: what is .cpp? | 11:24 |
wiking | http://snipt.org/uhhp6#expand | 11:25 |
wiking | looool | 11:25 |
wiking | niiice | 11:25 |
wiking | it's most def thinking that it's still an abstract class | 11:25 |
wiking | features/LatentFeatures.cpp:26:13: error: allocating an object of abstract class type 'CLatentFeatures' | 11:25 |
wiking | return new CLatentFeatures (*this); | 11:25 |
blackburn | wiking: but there is CLatentFeatures(CLatentFeatures) constructor? | 11:26 |
blackburn | where* | 11:26 |
wiking | i haven't defined a copy constructor | 11:26 |
wiking | mmm | 11:27 |
wiking | i think i see the problem | 11:27 |
wiking | ../shogun/features/Features.h:100:25: note: unimplemented pure virtual method 'get_feature_class' in 'CLatentFeatures' | 11:27 |
wiking | virtual EFeatureClass get_feature_class()=0; | 11:27 |
wiking | using inline is not good | 11:28 |
blackburn | wiking: ah | 11:28 |
blackburn | don't use inline on virtuals | 11:28 |
blackburn | makes no sense | 11:28 |
blackburn | as calling the virtual function leads to look up in vtable it can't be inlined afaik | 11:30 |
wiking | blackburn: have you tried this: git grep "inline virtual" | 11:30 |
blackburn | wiking: heh you are right | 11:30 |
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has joined #shogun | 11:31 | |
harshit_ | blackburn, : having problems with tsvm, please help ! | 11:31 |
wiking | ok this is getting funnier and funnier | 11:31 |
wiking | i still get the same error | 11:31 |
blackburn | wiking: still? | 11:31 |
wiking | yep | 11:31 |
wiking | even though i've moved the thing into .cpp | 11:31 |
wiking | for get_feature_class | 11:31 |
blackburn | harshit_: what is the problem? | 11:32 |
wiking | mmm | 11:32 |
wiking | i might know what's the problem | 11:32 |
wiking | ahahah yeah | 11:32 |
wiking | so can i have a question? | 11:32 |
wiking | why virtual EFeatureClass get_feature_class()=0; and not virtual EFeatureClass get_feature_class() const=0; | 11:32 |
wiking | ? | 11:32 |
harshit_ | blackburn: I am submitting correct parameters to ssl.cpp but getting wrong weights as output | 11:32 |
-!- xiangwang [~chatzilla@159.226.60.224] has joined #shogun | 11:32 | |
wiking | in CFeatures... | 11:33 |
xiangwang | hi wiking | 11:33 |
blackburn | wiking: const? yes this should be fixed | 11:33 |
wiking | xiangwang: heey | 11:33 |
wiking | blackburn: it's not const | 11:33 |
blackburn | wiking: I meant that | 11:33 |
wiking | i naturally assumed it should be a const as it does not change anything in the class.. | 11:33 |
blackburn | wiking: yes it should | 11:33 |
wiking | and that was the problem... since really it was a different func | 11:33 |
wiking | ok i'll write up a sed script for fixing this | 11:34 |
blackburn | hmm it seems I have a task for gsomix | 11:34 |
blackburn | gsomix: ! | 11:34 |
blackburn | :D | 11:34 |
-!- gsomix [~gsomix@188.168.4.185] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 11:35 | |
-!- gsomix [~gsomix@188.168.4.185] has joined #shogun | 11:35 | |
gsomix | blackburn, hey | 11:35 |
xiangwang | in SVMLight.h not all the functions of SVMLight realized ? like svm_learn_optimization? | 11:35 |
harshit_ | blackburn: please have a look at the comment I posted on my pull request ! | 11:36 |
blackburn | gsomix: you may fix the const issue | 11:36 |
blackburn | i.e. all the methods that should be const should be const | 11:36 |
gsomix | blackburn, ok/ | 11:37 |
blackburn | harshit_: checking | 11:37 |
blackburn | harshit_: ok I can't help with some code right now but you may compare our ssl.cpp and the one you assume as ground truth | 11:39 |
harshit_ | blackburn: ok no worries :) | 11:39 |
harshit_ | blackburn: and yeah I did compare that only .. but the problem was that there were a few functions which where not ported to shogun | 11:40 |
harshit_ | so I did that and when now I am comparing the 2 results it is not the same !!! | 11:40 |
blackburn | hmm that's strange | 11:41 |
blackburn | are you sure all the parameters are the same then? | 11:41 |
n4nd0 | harshit_: hey! did you compare the results of those funcitons you ported too? | 11:41 |
wiking | ? | 11:41 |
wiking | xiangwang: u wanted to msg me? | 11:41 |
blackburn | wiking: ? | 11:42 |
blackburn | ah | 11:42 |
blackburn | :D | 11:42 |
harshit_ | n4nd0: hey ! yeah ,Actually they are only 2 functions for creating subset of labeled features from mix of labeled and unlabeled | 11:42 |
xiangwang | sorry, wrong button:) | 11:42 |
harshit_ | that I ported | 11:42 |
harshit_ | n4nd0, and I think they are working fine | 11:42 |
wiking | xiangwang: ah ok :) | 11:42 |
wiking | blackburn: ehhehe of course now it compiles... ;) | 11:43 |
n4nd0 | harshit_: uhm, any clue where the error can be then? | 11:43 |
xiangwang | i'm trying to realize Latent SVM, but the code is not fully written. | 11:43 |
blackburn | wiking: I kind of delegated this task to gsomix so we will ensure that const is in place | 11:44 |
wiking | blackburn: actually virtual int32_t get_size (); should be the same | 11:44 |
wiking | blackburn: ahhaha :) | 11:44 |
harshit_ | n4nd0, somewhere in between ssl.cpp( file where svmlin code for shogun is ) | 11:44 |
blackburn | I have to go now guys :) n4nd0 thanks I hope you would help harshit_ ;) | 11:44 |
wiking | xiangwang: which code? | 11:45 |
xiangwang | should this part of code push to the shogun svm for gsoc qualification patch? | 11:45 |
n4nd0 | blackburn: I will try :) | 11:45 |
n4nd0 | harshit_: between ssl.cpp and? | 11:45 |
xiangwang | Intergration of Latent SVM | 11:45 |
xiangwang | to shogun | 11:45 |
harshit_ | n4nd0, I meant inside ssl.cpp | 11:45 |
wiking | xiangwang: i think you should be able to do a pull request... | 11:46 |
wiking | with some patch | 11:46 |
harshit_ | hey maybe i should test svm of svmlin | 11:46 |
xiangwang | but the code is not fully written... | 11:46 |
n4nd0 | harshit_: yeah, sure | 11:46 |
harshit_ | that would give me idea whether svmlin is correctly ported or not | 11:46 |
* blackburn is looking how one candidate student is asking his competitor how to apply to the task | 11:46 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 11:47 | |
xiangwang | ok I'll push for several times with the code being written. | 11:47 |
-!- xiangwang [~chatzilla@159.226.60.224] has quit [Quit: ChatZilla 0.9.88.2 [Firefox 11.0/20120312181643]] | 11:48 | |
harshit_ | n4nd0, btw is your covertree done ?? | 11:49 |
wiking | blackburn :DDD | 11:49 |
n4nd0 | harshit_: it is working and it is faster than what we used to have if the number of queries is more or less large | 11:49 |
n4nd0 | harshit_: but it is not yet as fast as the standalone version | 11:50 |
n4nd0 | harshit_: they use a computation of distance with quite a bit of speedups, I have still to work on those | 11:50 |
harshit_ | n4nd0, oh great , whats standalone version ? | 11:50 |
n4nd0 | harshit_: the original version from John Langford | 11:50 |
harshit_ | you mean actual implementation | 11:50 |
harshit_ | oh okay | 11:51 |
harshit_ | great dude :) | 11:51 |
harshit_ | Inform me when you'll get this merged, I can test it on my project (I did one project in which I used knn) | 11:52 |
n4nd0 | sure | 11:52 |
n4nd0 | that could help :) | 11:52 |
n4nd0 | what was the number of training vectors? | 11:52 |
n4nd0 | how many queries? | 11:52 |
n4nd0 | and what dimension? | 11:53 |
n4nd0 | lot of questions :)) | 11:53 |
harshit_ | really huge :) I used subset of 20 Newsgroup dataset | 11:53 |
harshit_ | i dont remember exactly but it was about 5000 training vectors and 4000 features/words | 11:53 |
harshit_ | That was a stupid project actually, I did that when I was first learning machine learning. | 11:54 |
harshit_ | n4nd0, And It used to take 5-6 hrs to get results on matlab ( I used my own implementation of knn) | 11:55 |
n4nd0 | harshit_: ok | 11:56 |
n4nd0 | harshit_: so feature vectors of dimension equal to 4000? that's a big one | 11:56 |
harshit_ | n4nd0, nlp tasks normally have bow representation of documents which make feature dimensions really large | 11:57 |
n4nd0 | ok | 11:58 |
harshit_ | n4nd0, I dont know if shogun has any support for sparse features ? | 11:58 |
harshit_ | do you knw if it has | 11:58 |
n4nd0 | yeah | 12:02 |
n4nd0 | CSparseFeatures | 12:02 |
n4nd0 | harshit_: https://github.com/iglesias/shogun/blob/master/src/shogun/features/SparseFeatures.h | 12:03 |
harshit_ | but that doesn't work in octave. right ? | 12:03 |
harshit_ | oh thats really good | 12:05 |
n4nd0 | I remember I heard something about that the swig typemaps were not ready for that | 12:09 |
n4nd0 | gsomix: I think you know something else about that ^, don't you? | 12:09 |
gsomix | n4nd0, mmm, it work in octave now. | 12:12 |
n4nd0 | cool | 12:12 |
n4nd0 | harshit_: then it looks like it is ready | 12:12 |
n4nd0 | gsomix: it was something you did recently? | 12:13 |
gsomix | n4nd0, yep. | 12:14 |
gsomix | harshit_, examples/undocumented/octave_modular/features_sparse_modular.m | 12:15 |
-!- abhishek [dce1fbc1@gateway/web/freenode/ip.220.225.251.193] has joined #shogun | 12:20 | |
-!- abhishek is now known as Guest58791 | 12:21 | |
-!- Guest58791 [dce1fbc1@gateway/web/freenode/ip.220.225.251.193] has left #shogun [] | 12:22 | |
-!- abhi1234 [dce1fbc1@gateway/web/freenode/ip.220.225.251.193] has joined #shogun | 12:22 | |
-!- cronor [~cronor@g231230224.adsl.alicedsl.de] has quit [Quit: cronor] | 12:28 | |
harshit_ | gsomix, n4nd0 : great,thats really helpful :) | 12:28 |
n4nd0 | no problem | 12:28 |
n4nd0 | I have to go now | 12:28 |
n4nd0 | see you guys! | 12:28 |
harshit_ | me too :) bye | 12:28 |
harshit_ | adios ! | 12:29 |
-!- harshit_ [~harshit@182.68.167.125] has quit [Quit: Leaving] | 12:29 | |
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has quit [Quit: Page closed] | 12:29 | |
wiking | abhi1234: yes? | 12:30 |
wiking | abhi1234: here | 12:30 |
abhi1234 | i have been using | 12:30 |
abhi1234 | hi wiking | 12:30 |
wiking | hey | 12:30 |
abhi1234 | i am sorry i was sending you personal messages | 12:31 |
wiking | nw | 12:31 |
abhi1234 | this cygwin thing is giving me a hard time | 12:31 |
wiking | oh i'm sorry | 12:31 |
abhi1234 | building modshogun.dll | 12:31 |
wiking | but i think i'm not the best person to ask about compiling shogun on windows :( | 12:31 |
abhi1234 | can you tell me one thing | 12:31 |
abhi1234 | cant we get a dll directly from the website | 12:32 |
wiking | abhi1234: it's better to ask here about it since i think somebody here might have better experience with shogun and windows | 12:32 |
abhi1234 | ohh | 12:32 |
abhi1234 | sure | 12:32 |
abhi1234 | thanks | 12:33 |
abhi1234 | can some one please tell me how to use shogun within visual studio environment | 12:33 |
abhi1234 | i am really having a hard time using libsvm dot net version | 12:34 |
wiking | abhi1234: and u could also try on the mailing list as some people from the shogun project are only checking the mailing list and are not online here | 12:38 |
abhi1234 | oh thanks | 12:38 |
abhi1234 | i will do that | 12:38 |
-!- xiangwang [~xiangwang@2001:cc0:2020:2013:f2de:f1ff:fe50:9012] has joined #shogun | 12:41 | |
wiking | xiangwang: please here | 13:04 |
* wiking thinks that he is becoming a git support line on #shogun | 13:06 | |
wiking | sonney2k: can u add this to the channel's topic: http://book.git-scm.com/ | 13:09 |
wiking | xiangwang: u should have your own forked git repo of shogun | 13:09 |
wiking | do the development there | 13:09 |
wiking | and commit/push all your changes there | 13:09 |
wiking | and once u have that | 13:09 |
wiking | u can do a pull request | 13:09 |
wiking | of course this is only true if u maintain your git repo on github | 13:10 |
CIA-64 | shogun: Michal Uricar master * r0f16159 / (2 files): Gauss-Seidel iterative method for Kernel Ridge Regression learning - http://git.io/69L0gg | 13:10 |
CIA-64 | shogun: Michal Uricar master * r7a4b0cc / (2 files): GS KRR work on comments - http://git.io/Va7eIA | 13:10 |
CIA-64 | shogun: Michal Uricar master * r4b4bde8 / (2 files): Indentation fixed. - http://git.io/L-xTXw | 13:10 |
CIA-64 | shogun: Soeren Sonnenburg master * r6653329 / (2 files): | 13:10 |
CIA-64 | shogun: Merge pull request #460 from uricamic/GSKRR | 13:10 |
CIA-64 | shogun: Gauss-Seidel iterative method for Kernel Ridge Regression learning - http://git.io/R0H15w | 13:10 |
wiking | yey | 13:10 |
wiking | uricamic: here? | 13:11 |
uricamic | wiking: yep | 13:15 |
wiking | uricamic: sorry for yesterday... :( | 13:18 |
uricamic | wiking: its ok | 13:18 |
wiking | do u have some time now? | 13:19 |
uricamic | wiking: yes, for now, but soon I will have to talk with my adviser | 13:20 |
xiangwang | thanks wiking, but i can't find my pull now in https://github.com/shogun-toolbox/shogun/pulls | 13:20 |
uricamic | wiking: but he is not here right now, so I have some time :) | 13:21 |
wiking | uricamic: ok so the thing is that instead of simply just adding 2 headers of libqp | 13:22 |
wiking | we should do some c++ wrapping of libqp | 13:22 |
uricamic | wiking: ok | 13:23 |
wiking | just like liblinear/libsvm or svmlight | 13:23 |
wiking | btw the libqp_splx_solver function in shogun case | 13:24 |
wiking | it's copied from libqp? | 13:24 |
wiking | ah yeah it is | 13:25 |
-!- PhilTillet [~Philippe@157.159.42.154] has joined #shogun | 13:25 | |
uricamic | wiking: I think that lipqp.h is the header file of libqp_splx.cpp, where is this solver implemented | 13:26 |
wiking | yep | 13:26 |
wiking | i was just wondering | 13:26 |
wiking | whether these are like copied from the libqp | 13:26 |
wiking | and actually it's just copied from libqp into shogun's repo | 13:27 |
wiking | and that's it basically | 13:27 |
uricamic | yep | 13:27 |
wiking | and actually at this moment only libqp_splx_solver is being copied | 13:27 |
wiking | so i guess we should follow the style of shogun, i.e. copy the source from libqp and not 'depend' on it ;) | 13:28 |
wiking | i have objections about this but then again i guess sonney2k and blackburn has their reasons to do this this way... | 13:28 |
wiking | but still we should do a c++ wrapper for both of the functions | 13:29 |
uricamic | by libqp you mean the original library http://cmp.felk.cvut.cz/~xfrancv/libqp/html/ ? | 13:29 |
wiking | yes | 13:29 |
uricamic | ok | 13:29 |
wiking | i don't know though how often vojtech is planning to do upgrades on libqp ;) | 13:30 |
wiking | from an email of his he seems to be quite busy lately | 13:30 |
wiking | so i guess not too soon | 13:30 |
uricamic | wiking: yes he is, I can ask him | 13:31 |
uricamic | but he is not here right now :) | 13:31 |
wiking | heheh his your advisor? | 13:31 |
wiking | *he's | 13:31 |
uricamic | yep :) | 13:32 |
wiking | ah cool! you are lucky! ;) | 13:32 |
uricamic | :) | 13:32 |
uricamic | but I don't think, that he plans to do some chenges | 13:33 |
wiking | ok | 13:35 |
wiking | so anyhow i'm just checking now where could we fit nicely the whole thing | 13:35 |
wiking | so whether we could fit it into one abstract class | 13:37 |
wiking | or simply just create an own standalone class for it | 13:38 |
wiking | sonney2k: here? | 13:41 |
wiking | afais we should just do a standalone cpp wrapper for it but i need to see what sonney2k thinks about it | 13:43 |
wiking | uricamic: r u going to continue bmrm? | 13:44 |
uricamic | wiking: I hope so | 13:45 |
uricamic | now it is basically in the state, that it also requires a wrapper | 13:46 |
-!- gsomix [~gsomix@188.168.4.185] has quit [Read error: Operation timed out] | 13:46 | |
sonne|work | wiking: whats up? | 13:46 |
Marty28 | cya | 13:48 |
wiking | sonne|work: it's just a quick question about libqp integration | 13:49 |
sonne|work | wiking: so are uricamic and you intending to do sth about it? | 13:50 |
wiking | sonne|work: yep hoping so | 13:50 |
sonne|work | would be great | 13:50 |
sonne|work | I assume you both need it for your projects anyways... | 13:50 |
uricamic | sonne|work: yep, I will definitely need it | 13:51 |
wiking | i was just wondering if u have any opinion about whether the libqp wrapper should be a standalone class (as afaik there's no abstract class where we could plug this into) or shall we start thinking about making an abstract qp solver class ...? i.e. maybe the latter one would be wiser as i can see there's the QPBSVMLib | 13:51 |
wiking | that i *guess* it's another qp solver | 13:52 |
-!- Marina [55b326f1@gateway/web/freenode/ip.85.179.38.241] has quit [Quit: Page closed] | 13:53 | |
wiking | so maybe having like a QP solver abstract class and doing the various implementations inherited from that would be good to have... | 13:54 |
wiking | uricamic: me too ;) | 13:54 |
sonne|work | wiking: I think it should be similar to what libocas is now - basically a C library we have in shogun and use with all the necessary callbacks to enable sufficient flexibility | 13:54 |
sonne|work | then a wrapper on top | 13:54 |
sonne|work | yes | 13:54 |
abhi1234 | hi sonne | 13:55 |
abhi1234 | i have a quick question | 13:55 |
abhi1234 | if it is a good time to ask | 13:55 |
sonne|work | this way vojtech can continue to maintain this library outside and we only have to adjust the wrapper | 13:55 |
abhi1234 | thanks in advance | 13:55 |
sonne|work | abhi1234: ask | 13:55 |
sonne|work | (don't ask to ask) | 13:55 |
abhi1234 | ohh | 13:55 |
abhi1234 | sorry | 13:55 |
wiking | sonne|work: mmmm | 13:55 |
abhi1234 | I am planning to run shogun in visual studio | 13:56 |
abhi1234 | is it possible | 13:56 |
abhi1234 | i have window 7 and cygwin | 13:56 |
abhi1234 | and it is really difficult to compile the modshogun.dll | 13:56 |
abhi1234 | in cygwin | 13:56 |
abhi1234 | so if it is possible can I get a modshogun.dll for a quick run | 13:57 |
sonne|work | wiking: what does that mean? | 13:57 |
wiking | sonne|work: just thinking loudly ;) | 13:57 |
sonne|work | abhi1234: I am afraid that this is not so easy | 13:58 |
sonne|work | no one has succeeded in a pure windows port of libshogun | 13:58 |
-!- xiangwang [~xiangwang@2001:cc0:2020:2013:f2de:f1ff:fe50:9012] has quit [Remote host closed the connection] | 13:58 | |
wiking | sonne|work: so let's keep the libqp as it is now (i.e. an easy way to add the future changes), but still aim for not only a nice c++ wrapper on top of the qp solver but start defining an abstract qp solver as well...? | 13:59 |
abhi1234 | ohh terrible then | 13:59 |
abhi1234 | i am really impressed by the kind of algorithm available in shogun | 14:00 |
abhi1234 | you people have done an amazing work | 14:00 |
wiking | abhi1234: maybe u could be the one who does the work ;) | 14:00 |
abhi1234 | yeah i would like to be | 14:00 |
wiking | abhi1234: what's your current error?: ) | 14:00 |
sonne|work | wiking: that is what I meant - some nice C++ interface to enable using it within shogun and all its interfaces | 14:00 |
sonne|work | wiking: but using libqp internally | 14:01 |
abhi1234 | my current error is related to option -pthread | 14:01 |
wiking | sonne|work: ok... and try to do something about QPBSVMLib as well... ;) | 14:01 |
wiking | sonne|work: ooh i think we are on the same page then... i saw blackburn's possible patch to move out to an 'external' directory the external libraries... i totally second that idea | 14:02 |
abhi1234 | but cant i get the modshogun.dll from someone who built it in linux environment only | 14:02 |
wiking | abhi1234: well it's not a dll | 14:02 |
wiking | i can give u a .so | 14:02 |
wiking | but i doubt that it'll work for u | 14:02 |
abhi1234 | ohh | 14:02 |
sonne|work | wiking: well if we manage that kind of integration, the files cplex / pr_loqo etc should all be in there too | 14:04 |
sonne|work | and e.g. SVMlight could work with any of the solvers then | 14:04 |
abhi1234 | .so i am afraid wont work for me | 14:04 |
wiking | sonne|work: cool | 14:04 |
wiking | i see my competitor did a pull request i should get onto giving a patch today/tomorrow ;P | 14:04 |
wiking | abhi1234: ok so you die with -pthreads | 14:05 |
wiking | u use cygwin i assume | 14:05 |
wiking | abhi1234: have u actually 'made' cygwin to install the pthread library? | 14:06 |
wiking | as it is part of cygwin but you need to tell the installer that u want it | 14:06 |
wiking | if not then try to install it now | 14:06 |
abhi1234 | sure | 14:06 |
abhi1234 | let me try it | 14:06 |
wiking | mmm | 14:06 |
wiking | can u do this | 14:06 |
wiking | ls -la /usr/lib/libpth* | 14:06 |
abhi1234 | one moment | 14:07 |
wiking | and what's the result for it | 14:07 |
wiking | uricamic: u still here or your advisor caught you? | 14:08 |
abhi1234 | -rw-r--r-- 1 | 14:08 |
wiking | abhi1234: the whole lien ;) | 14:08 |
abhi1234 | is the result i get | 14:08 |
uricamic | wiking: still here :) | 14:08 |
wiking | *line | 14:08 |
abhi1234 | one moment | 14:08 |
wiking | uricamic: so i think we should then start defining an abstract qp solver class | 14:08 |
uricamic | wiking: yep | 14:09 |
abhi1234 | -rw-r--r-- 1 TTG None 30678 Apr 5 20:35 /usr/lib/libpthread.a | 14:09 |
wiking | abhi1234: so u have then actually pthread | 14:10 |
wiking | abhi1234: why don't you pastebin your full error and send it to us | 14:10 |
wiking | i mean compilation error | 14:10 |
abhi1234 | sure I will do that | 14:11 |
abhi1234 | one moment | 14:11 |
wiking | uricamic: ok i'll sketch up something soon and put it into one branch of mine so that everybody can comment on it via github, but if you have already some ideas please don't be shy ;) | 14:11 |
uricamic | wiking: ok, I will look at it | 14:12 |
abhi1234 | earlier when i tried to compile it was throwing error "unrecognized option -pthread" | 14:15 |
abhi1234 | now it is showing the same thing as warning | 14:16 |
-!- pluskid [~pluskid@117.136.17.61] has joined #shogun | 14:30 | |
-!- gsomix [~gsomix@188.168.2.191] has joined #shogun | 14:33 | |
-!- abhi1234 [dce1fbc1@gateway/web/freenode/ip.220.225.251.193] has quit [Ping timeout: 245 seconds] | 14:37 | |
PhilTillet | hello everybody :) | 14:41 |
Marty28 | Does anyone know the ERROR "ImportError: libshogun.so.12: cannot open shared object file: No such file or directory"? I just installed a shogun version locally and this comes upon importing shogun. | 14:41 |
Marty28 | I mean import modshogun. | 14:41 |
Marty28 | (trying to use another destdir) | 14:44 |
-!- cronor [~cronor@fb.ml.tu-berlin.de] has joined #shogun | 14:56 | |
pluskid | which distribution are you using? and how did you install Shogun? | 15:00 |
Marty28 | might be NTFS | 15:09 |
Marty28 | cannot change rights on this drive | 15:09 |
Marty28 | ok, moving to /HOME/somewhere | 15:09 |
Marty28 | the shogun_modular sonne buildt yesterday | 15:11 |
Marty28 | -d | 15:12 |
-!- pluskid [~pluskid@117.136.17.61] has quit [Ping timeout: 265 seconds] | 15:22 | |
Marty28 | nah, does not work | 15:33 |
Marty28 | hiho | 15:56 |
Marty28 | I did the following: | 15:57 |
Marty28 | ./configure --enable-svm-light --python=python2.7 --destdir=/home/ralf/shogun-builds --prefix=/ | tee configure_log.txt | 15:57 |
Marty28 | make | tee makelog.txt | 15:57 |
Marty28 | sudo make install | tee makeinst_log.txt | 15:57 |
Marty28 | THEN: ~/shogun-builds/shogun0417/bin$ shogun | 15:58 |
Marty28 | shogun: error while loading shared libraries: libshogun.so.12: cannot open shared object file: No such file or director | 15:58 |
Marty28 | Why cannot the libshogun.so.12 be found? | 15:58 |
Marty28 | Is it because of <configure_log.txt> "Checking for libshogun ... yes (internal in source tree)" ? | 15:59 |
Marty28 | I already installed shogun 1.10 and now I am installing the current version from github | 16:00 |
sonne|work | Marty28: you need to set you LD_LIBRARY_PATH | 16:01 |
sonne|work | to the dir where libshogun*.so* is | 16:01 |
Marty28 | ok | 16:02 |
Marty28 | shell variable? | 16:02 |
Marty28 | OIC | 16:03 |
Marty28 | BTW: http://www.shogun-toolbox.org/doc/en/current/installation.html contains outdated interfaces, right? | 16:03 |
-!- PhilTillet [~Philippe@157.159.42.154] has quit [Read error: Operation timed out] | 16:08 | |
Marty28 | whew, works | 16:10 |
Marty28 | thanks | 16:10 |
Marty28 | sonne: git clone git://github.com/shogun-toolbox/shogun.git | 16:17 |
Marty28 | sonne: Do I get the latest shogun version with this including the changes from yesterday? | 16:17 |
sonne|work | yes | 16:19 |
Marty28 | Did you change WeightedDegreePositionStringKernel.prepare_POIM2 to SGVector/Matrix? | 16:22 |
sonne|work | yes | 16:24 |
sonne|work | wait | 16:24 |
sonne|work | nope | 16:24 |
sonne|work | will do it on the train ride then | 16:24 |
sonne|work | gtg | 16:24 |
sonne|work | cu | 16:24 |
Marty28 | super :) | 16:24 |
Marty28 | biba | 16:25 |
-!- vikram360 [~vikram360@117.192.185.94] has quit [Ping timeout: 260 seconds] | 16:27 | |
-!- vikram360 [~vikram360@117.192.163.173] has joined #shogun | 16:28 | |
@sonney2k | wiking, btw have you tried to create dotfeatures from your homo kernel map? | 16:44 |
-!- Marina__ [8d175a52@gateway/web/freenode/ip.141.23.90.82] has joined #shogun | 16:45 | |
Marina__ | sonney2k: i tried again... but it doesn't work | 16:46 |
Marina__ | i get the same error | 16:46 |
Marina__ | i have install the latest version of shogun | 16:46 |
Marina__ | and i still get the error: | 16:46 |
Marina__ | File "/home/mvidovic/POIM/poim.py", line 37, in compute_poims kernel.prepare_POIM2(distr,int(poimdegree), int(max_len)) TypeError: in method 'WeightedDegreePositionStringKernel_prepare_POIM2', argument 2 of type 'float64_t *' | 16:46 |
CIA-64 | shogun: Soeren Sonnenburg master * r9067389 / (2 files): convert prepare_POIM2 to SGMatrix - http://git.io/SFvtEA | 16:47 |
@sonney2k | Marina__, voila | 16:47 |
Marina__ | so i have to go to src/shogun/kernel/WeightedDegreePositionStringKernel.cpp and delete and add the lines?? | 16:49 |
CIA-64 | shogun: Evgeniy Andreev master * r46fd9e4 / (67 files in 2 dirs): fixed string list functions - http://git.io/KuM9fw | 16:49 |
CIA-64 | shogun: Soeren Sonnenburg master * ra8231f2 / (67 files in 2 dirs): | 16:49 |
CIA-64 | shogun: Merge pull request #466 from gsomix/python3_interface | 16:49 |
CIA-64 | shogun: Python 3 support for static python interface - http://git.io/c4UuhA | 16:49 |
@sonney2k | Marina__, or checkout the whole beast | 16:49 |
@sonney2k | git pull | 16:50 |
@sonney2k | Marty28, done | 16:50 |
@sonney2k | needs someone doing the testing though | 16:50 |
Marina__ | @sonney2k: I checkout the whole src - isn't that enough? | 16:52 |
@sonney2k | Marina__, you do git clone once and then only update via git pull | 16:54 |
@sonney2k | no need to checkout the whole source | 16:54 |
@sonney2k | again and again | 16:54 |
Marty28 | thx | 16:56 |
* sonney2k leaves the train | 16:59 | |
@sonney2k | cu | 16:59 |
Marty28 | cya | 17:01 |
Marty28 | Is the git pull syntax "git pull git://github.com/shogun-toolbox/shogun.git"? | 17:06 |
Marty28 | hmmm | 17:09 |
Marina__ | @sonney: it works :D | 17:10 |
Marty28 | sonne: Works here as well | 17:14 |
Marty28 | Now the result has to make sense as well - lets see... | 17:15 |
gsomix | sonney2k, thanks. :) | 17:16 |
-!- blackburn [~qdrgsm@188.168.2.179] has joined #shogun | 17:18 | |
-!- Marty28 [~Marty@158.181.78.199] has quit [Quit: ChatZilla 0.9.88.2 [Firefox 11.0/20120310010446]] | 17:22 | |
blackburn | wiking: I just read logs and :D :D | 17:23 |
-!- uricamic [9320543b@gateway/web/freenode/ip.147.32.84.59] has quit [Quit: Page closed] | 17:44 | |
-!- cronor_ [~cronor@141.23.80.206] has joined #shogun | 18:20 | |
-!- cronor [~cronor@fb.ml.tu-berlin.de] has quit [Read error: Operation timed out] | 18:22 | |
-!- cronor_ [~cronor@141.23.80.206] has quit [Ping timeout: 276 seconds] | 18:25 | |
-!- cronor [~cronor@141.23.80.206] has joined #shogun | 18:26 | |
-!- cronor [~cronor@141.23.80.206] has quit [Remote host closed the connection] | 18:26 | |
-!- cronor [~cronor@fb.ml.tu-berlin.de] has joined #shogun | 18:27 | |
-!- cronor [~cronor@fb.ml.tu-berlin.de] has quit [Client Quit] | 18:30 | |
@sonney2k | Marina__, let us know if the results are OK | 18:57 |
-!- blackburn [~qdrgsm@188.168.2.179] has quit [Quit: Leaving.] | 18:59 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 19:08 | |
@sonney2k | gsomix, keep them coming | 19:10 |
@sonney2k | gsomix, which reminds me are you progressing with sgvector stuff? | 19:10 |
n4nd0 | sonney2k: I apologize that I didn't change the time thing in KNN, I forgot about it :S | 19:11 |
n4nd0 | sonney2k: I will apply the change now | 19:11 |
@sonney2k | n4nd0, great to hear | 19:12 |
@sonney2k | this thing should finally be merged | 19:12 |
n4nd0 | sonney2k: yeah, my pace has been reduced a bit, exams coming next week! :O | 19:13 |
@sonney2k | n4nd0, no problem | 19:13 |
@sonney2k | btw did anyone apply at some other org? | 19:13 |
n4nd0 | I didn't | 19:14 |
n4nd0 | sonney2k: should I have done it? :P | 19:16 |
@sonney2k | hehe | 19:17 |
@sonney2k | apr 23 is the day you will know | 19:17 |
n4nd0 | :) | 19:17 |
n4nd0 | sonney2k: by the way, about that meeting you were talking about, the one where the timezone was an issue | 19:18 |
n4nd0 | that is something just for mentors right? | 19:18 |
Marina__ | @sonney2k: my results are ok - they are like the results half a year ago! | 19:18 |
@sonney2k | n4nd0, no students + mentors | 19:18 |
@sonney2k | Marina__, good to hear | 19:19 |
n4nd0 | Marina__: awesome! | 19:19 |
@sonney2k | so then happy poim'ing | 19:19 |
wiking | blackburn what? :) | 19:19 |
wiking | sonney2k: noun not yet... i think first the homo kernel map should be extended with the RBF approximation at least | 19:20 |
Marina__ | :D | 19:22 |
@sonney2k | wiking, but that is just an addon | 19:32 |
@sonney2k | n4nd0, btw I guess you didn't have time to compare things with the dnrm2? | 19:32 |
@sonney2k | cblas function | 19:32 |
@sonney2k | for computing the euclidean distance? | 19:32 |
n4nd0 | no, I have not done that yet | 19:33 |
n4nd0 | I will start looking into that once the current patch is merged | 19:33 |
* sonney2k thinks we should add a norm function to DotFeatures | 19:33 | |
@sonney2k | n4nd0, would be nice - but first things first | 19:34 |
n4nd0 | sonney2k: your idea was that it should be faster with that than doing our own version of JL's distance function | 19:34 |
n4nd0 | with unrolled loops and so on | 19:34 |
@sonney2k | n4nd0, yeah | 19:35 |
@sonney2k | and screw the (small) speedup that the upper bound to distance gives us | 19:36 |
n4nd0 | ok | 19:36 |
-!- Marina__ [8d175a52@gateway/web/freenode/ip.141.23.90.82] has quit [Quit: Page closed] | 19:46 | |
-!- Marty28_ [~marty@158.181.78.199] has joined #shogun | 19:52 | |
Marty28_ | test | 19:52 |
@sonney2k | ? | 19:57 |
Marty28_ | New irc client on ipad :) | 20:06 |
Marty28_ | My testing of poim with artificial data is set to tomorrow. | 20:08 |
-!- Marty28_ [~marty@158.181.78.199] has quit [Quit: Marty28_] | 20:09 | |
@sonney2k | n4nd0, do you have a ptr to JL's distance code | 20:20 |
@sonney2k | the snipped you pasted? | 20:20 |
n4nd0 | sonney2k: yeah, give me a sec | 20:22 |
* gsomix have returned from dance floor. | 20:24 | |
n4nd0 | sonney2k: http://snipt.org/uhhP1#expand there it goes | 20:25 |
n4nd0 | sonney2k: do we need the assumption of length multiple of 8 in order to use cblas? | 20:25 |
n4nd0 | I guess not | 20:25 |
gsomix | sonney2k, I'm doing SGVector -> SGVector& conversion for the arguments. | 20:27 |
gsomix | sonney2k, slowly at this days. trying to deal with study in university and with choosing supervisor. | 20:27 |
-!- vikram360 [~vikram360@117.192.163.173] has quit [Read error: Connection reset by peer] | 20:31 | |
-!- vikram360 [~vikram360@117.192.163.173] has joined #shogun | 20:32 | |
@sonney2k | gsomix, ok just keep me updated... | 20:35 |
@sonney2k | n4nd0, not necessary | 20:35 |
-!- harshit_ [~harshit@182.68.167.125] has joined #shogun | 20:37 | |
gsomix | sonney2k, ok. | 20:43 |
@sonney2k | n4nd0, I think that JL's code can be twice as fast if he uses d1 ...d4 | 20:51 |
@sonney2k | then it would really be just two sse instructions | 20:51 |
@sonney2k | but I guess we need a separate benmark for that | 20:51 |
n4nd0 | sonney2k: yeah that makes actually sense, I thought of it, if they unrolled 2, why not 4? but I thought that maybe it didn't pay off or sth | 20:52 |
@sonney2k | maybe when he wrote it it didn't make any difference | 20:52 |
@sonney2k | or he wrote it for double | 20:52 |
@sonney2k | n4nd0, still I don't understand why you measured that in JL's implementation the thing is *much* fuster | 20:53 |
@sonney2k | faster | 20:53 |
@sonney2k | (like 10x or so) | 20:53 |
n4nd0 | do you think it shouldn't? | 20:53 |
@sonney2k | this thing here looks like it can be twice as fast due to the d1,d2 idea and another factor of 2 due to using floats | 20:54 |
@sonney2k | so 4x | 20:54 |
@sonney2k | but no more | 20:54 |
@sonney2k | (if the upper bound doesn't help a lot) | 20:54 |
n4nd0 | some overhead in shogun somewhere? | 20:55 |
n4nd0 | I will profile this in any case, I find it interesting | 20:55 |
@sonney2k | n4nd0, I would start with some simple example just computing the distance matrix for say 1000 vectors once with the shogun and once with the jl code | 20:55 |
@sonney2k | I would predict 4 times faster :D | 20:56 |
n4nd0 | I write that down, we'll see if your prediction is correct ;) | 20:56 |
@sonney2k | n4nd0, then you could use the *slow* shogun distance version in Jl's tree and see if speeds are the same | 20:59 |
@sonney2k | if not then we have a timesink somewhere :) | 20:59 |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 21:02 | |
gsomix | it's time to reading the philosophy. | 21:13 |
gsomix | cu, guys | 21:13 |
@sonney2k | n4nd0, actually I bet on the time sink - the distance function will only be marginally faster | 21:17 |
gsomix | "Machine Learning Andrew Ng, Associate Professor 23 April 2012" wow, most interesting things are after 23 :) | 21:34 |
n4nd0 | sonney2k: where the time sink would be then? | 21:35 |
@sonney2k | n4nd0, that is the hard part to find out :) | 21:37 |
@sonney2k | n4nd0, but if this distance function is soo fast it would speed up the gaussian kernel computation and lots of stuff *a lot* | 21:37 |
@sonney2k | n4nd0, which reminds me that we can speed things up anyways by using ||x-y||^2 == x^T x - 2xy + y^Ty | 21:39 |
@sonney2k | x^Tx can be precomputed | 21:39 |
n4nd0 | sonney2k: would that entail a big difference in general? | 21:40 |
@sonney2k | and so we only need dot(x,y) | 21:40 |
@sonney2k | that is what we use for gaussian kernel at least for sparse features | 21:40 |
@sonney2k | actually for all features | 21:41 |
@sonney2k | that certainly is *a lot* faster even than what JL uses | 21:42 |
@sonney2k | n4nd0, so yes :D | 21:43 |
@sonney2k | n4nd0, I hope you pass exams soon and have some time to investigate | 21:43 |
n4nd0 | sonney2k: there was also another speedup suggested in a comment within his code but not implemented | 21:48 |
n4nd0 | sonney2k: https://github.com/iglesias/shogun/blob/patch-jlcovertree/src/shogun/lib/JLCoverTree.h | 21:49 |
n4nd0 | line 594 | 21:49 |
@sonney2k | no idea about that (no covertree guru) | 21:51 |
n4nd0 | sonney2k: I have updated the patch with time functions used from shogun | 22:03 |
@sonney2k | then let me merge it | 22:03 |
n4nd0 | sonney2k: can you please take it a look and tell me if I have done the git part correctly? I see that this time there are no duplicates commits by me | 22:03 |
n4nd0 | but commits from other people appear in the PR :S | 22:04 |
@sonney2k | n4nd0, you did it again | 22:04 |
@sonney2k | no git pull --rebase | 22:04 |
n4nd0 | I did that this time! | 22:04 |
@sonney2k | why am I seeing https://github.com/shogun-toolbox/shogun/pull/464 | 22:04 |
@sonney2k | uricamic and others added some commits 2 days ago | 22:04 |
@sonney2k | 3edd188 | 22:04 |
@sonney2k | Gauss-Seidel iterative method for Kernel Ridge Regression learning | 22:04 |
@sonney2k | ... | 22:04 |
n4nd0 | yeah I know but I did it that way this time | 22:05 |
@sonney2k | https://github.com/shogun-toolbox/shogun/pull/464/commits | 22:05 |
@sonney2k | n4nd0, can you tell me exactly what you did? | 22:05 |
n4nd0 | http://snipt.org/uhhT8#expand | 22:05 |
n4nd0 | that's my history | 22:05 |
n4nd0 | grepped to git commands | 22:05 |
n4nd0 | let's find out what I am doing wrong, I don't really understand :( | 22:06 |
n4nd0 | the experimental stuff is just something I did to try out if I got in that case the non-fast-forward allowed and how I can delete a remote branch | 22:07 |
n4nd0 | 2058 gave the familiar error about non-fast-forward not allowed | 22:08 |
n4nd0 | so then I git pull --rebase | 22:08 |
@sonney2k | might be related to 2060 git pull --rebase | 22:10 |
@sonney2k | 2061 git pull --rebase origin patch-jlcovertree | 22:10 |
@sonney2k | I mean *I* would have git checkout master ; then git pull --rebase | 22:10 |
@sonney2k | then git checkout patch-jlcovertree | 22:10 |
@sonney2k | git rebase master | 22:10 |
@sonney2k | git push origin patch-jlcovertree | 22:11 |
n4nd0 | even if the problem is to push in the branch, do I have to pull in my master? | 22:11 |
n4nd0 | as you see in 2040, the push in master worked fine | 22:11 |
@sonney2k | did you do work inbetween 2040 and 2060? | 22:12 |
@sonney2k | anyway I think the problem is the call to git pull --rebase origin patch-jlcovertree | 22:13 |
@sonney2k | I think this rebases everything new on top of what is in origin patch-jlcovertree | 22:14 |
n4nd0 | between 2040 and 2060 I didn't work | 22:14 |
n4nd0 | the whole history is there | 22:14 |
@sonney2k | I think you don't ever need to pull from that and if you do not with rebase | 22:15 |
n4nd0 | :O | 22:16 |
n4nd0 | I thought you told me that last time | 22:16 |
n4nd0 | so then, do you have any clue about how should I do it? | 22:17 |
@sonney2k | no I said you shoudl do git checkout master; git pull --rebase ; git checkout patch-jlcovertree ; git rebase master ; git push origin patch-jlcovertree | 22:17 |
@sonney2k | n4nd0, do you still have the patch lying around isolated? | 22:18 |
@sonney2k | or can you follow what is described in git help reset | 22:19 |
@sonney2k | and in there 'Undo a merge or pull inside a dirty working tree' | 22:19 |
n4nd0 | I can do as I did for this pull request, create a new branch and apply the patches for each of the commits | 22:19 |
@sonney2k | ok | 22:19 |
wiking | nyipp | 22:24 |
-!- ckwidmer [~chris@HSI-KBW-046-005-237-106.hsi8.kabel-badenwuerttemberg.de] has joined #shogun | 22:32 | |
wiking | sonney2k: around? | 22:32 |
wiking | or anybody has been doing some hierarchical classification lately? | 22:32 |
@sonney2k | wiking, topic hopping again? | 22:37 |
n4nd0 | sonney2k: there it goes the new PR and the other closed | 22:37 |
n4nd0 | sonney2k: I am sorry for the inconveniences due to my git problems :S | 22:38 |
CIA-64 | shogun: Soeren Sonnenburg master * r3d4d21f / (8 files in 3 dirs): | 22:38 |
CIA-64 | shogun: Merge pull request #467 from iglesias/repath-jlcovertree | 22:38 |
CIA-64 | shogun: John Langford's Cover Tree Integration (+8 more commits...) - http://git.io/MXkc1w | 22:38 |
n4nd0 | it looks that I'll learn it the hard way | 22:38 |
n4nd0 | :( | 22:38 |
@sonney2k | n4nd0, don't worry | 22:38 |
wiking | sonney2k: nah ... just i'll need it soonish | 22:39 |
@sonney2k | last year we ourselves had lots of commits like this | 22:39 |
@sonney2k | and yes me and blackburn producing them too | 22:39 |
wiking | sonney2k: since now i'm doing a classification challenge and today they've released the training set... and when i saw it i've realized that there's no fucking way that the task can be done without some hierarchy | 22:39 |
wiking | so i'm just 'preparing' myself for the future :))) | 22:40 |
wiking | hahahah today new spam day... getting 'part time job offers' :D | 22:41 |
-!- Marty28 [~chatzilla@158.181.78.199] has joined #shogun | 22:44 | |
@sonney2k | wiking, I see - and well it is part of the multiclass task we have this GSoC - some tree based ecoc... | 22:45 |
@sonney2k | anyways | 22:46 |
@sonney2k | nite everyone | 22:46 |
wiking | gnite | 22:46 |
gsomix | sonney2k, nite | 22:46 |
n4nd0 | sonney2k: good night | 22:46 |
wiking | n4nd0: yo | 22:46 |
n4nd0 | wiking: hey! | 22:46 |
wiking | mmm i think it's pluskid who's doing ^that | 22:47 |
n4nd0 | what? | 22:48 |
wiking | "well it is part of the multiclass task we have this GSoC - some tree based ecoc" | 22:48 |
n4nd0 | aham, who knows! | 22:48 |
wiking | hahahaha | 22:48 |
wiking | i guess soeren knows :> | 22:48 |
n4nd0 | :D | 22:50 |
-!- Marty28 [~chatzilla@158.181.78.199] has quit [Quit: ChatZilla 0.9.88.2 [Firefox 11.0/20120312181643]] | 22:51 | |
gsomix | good night | 22:52 |
n4nd0 | good night gsomix | 22:52 |
gsomix | philosophy puts to sleep =____= | 22:53 |
n4nd0 | wow | 22:55 |
n4nd0 | gsomix: philosophy exam? | 22:55 |
gsomix | n4nd0, just seminar. exam will be in june | 22:56 |
n4nd0 | for the philosophy exams tend to be like ... no idea what I am doing, sometimes I even laugh about it :D | 22:58 |
-!- harshit_ [~harshit@182.68.167.125] has quit [Remote host closed the connection] | 23:19 | |
-!- ckwidmer [~chris@HSI-KBW-046-005-237-106.hsi8.kabel-badenwuerttemberg.de] has quit [Remote host closed the connection] | 23:28 | |
-!- blackburn [~qdrgsm@188.168.2.179] has joined #shogun | 23:29 | |
blackburn | hey there | 23:30 |
n4nd0 | blackburn: hey! | 23:31 |
blackburn | damn I missed chris | 23:36 |
blackburn | sonney2k: still there? | 23:39 |
blackburn | n4nd0: are you finished with that stuff? | 23:40 |
n4nd0 | blackburn: with the cover tree pr? | 23:41 |
blackburn | yeap | 23:41 |
n4nd0 | yeah | 23:41 |
n4nd0 | the first version at least :) | 23:41 |
n4nd0 | now I am fighting agains my databases course | 23:41 |
blackburn | uh I see | 23:42 |
n4nd0 | literally, fighting hehe | 23:42 |
blackburn | when are you going to pass it? | 23:42 |
blackburn | gsomix: still here? | 23:42 |
n4nd0 | I am going to take the exam next Thursday | 23:42 |
n4nd0 | not tomorrow | 23:42 |
n4nd0 | in the next week | 23:42 |
n4nd0 | I am going to Spain on Tuesday :S | 23:42 |
blackburn | wtf | 23:42 |
blackburn | :D | 23:42 |
blackburn | are you going to take this exam in spain? | 23:43 |
n4nd0 | yeah | 23:43 |
n4nd0 | this one, and other two | 23:43 |
blackburn | will you go back to sweden then? | 23:44 |
n4nd0 | of course! | 23:44 |
n4nd0 | hehe, as soon as I am done with my last exam this period | 23:44 |
blackburn | I have to take exam today but I'm going to take it tomorrow as it also possible | 23:46 |
blackburn | and yeah I don't know anything yet | 23:47 |
blackburn | :D | 23:47 |
blackburn | n4nd0: when will you have last exam then? | 23:52 |
n4nd0 | blackburn: May 7th | 23:52 |
blackburn | just like me | 23:52 |
n4nd0 | blackburn: I've to reboot to windows :O brb | 23:52 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 23:52 | |
blackburn | (but 8) | 23:52 |
-!- n4nd0 [53b32c87@gateway/web/freenode/ip.83.179.44.135] has joined #shogun | 23:55 | |
n4nd0 | blackburn: so what is your exam about? | 23:55 |
blackburn | n4nd0: tomorrow's is called as 'optical informatics' | 23:56 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 23:56 | |
blackburn | a lot of shit there | 23:56 |
n4nd0 | blackburn: what is that about? | 23:57 |
n4nd0 | optics or informatics :P? | 23:57 |
blackburn | n4nd0: yes exactly they are all crazy | 23:57 |
blackburn | optics mainly | 23:57 |
blackburn | I bet they put some info* word there just to be it more official | 23:57 |
n4nd0 | aham | 23:57 |
n4nd0 | stuff like Fresnel coefficients, Snell law, reflexion and so on? | 23:58 |
blackburn | yes but more advanced | 23:58 |
blackburn | some photo stuff also | 23:58 |
n4nd0 | oh | 23:58 |
blackburn | abberations | 23:58 |
n4nd0 | I don't know about that | 23:58 |
blackburn | ccd matrices | 23:58 |
blackburn | me neither | 23:58 |
blackburn | next exams are on some software engineering, digital signal processing and (sic) politology | 23:59 |
--- Log closed Thu Apr 19 00:00:19 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!