--- Log opened Thu Apr 05 00:00:19 2012 | ||
gsomix | sonney2k, there is bug in strings. i already fixed it, but not posted code. | 00:00 |
---|---|---|
gsomix | sonney2k, 800MB. gcc 4.5.2 | 00:00 |
gsomix | strange | 00:00 |
@sonney2k | ? | 00:00 |
@sonney2k | how that? | 00:00 |
gsomix | sonney2k, are you asking about python3 or memory? | 00:02 |
gsomix | *or about memory | 00:02 |
gsomix | sonney2k, can you see my proposal? | 00:09 |
gsomix | at that time I can make corrections in python3 interface. | 00:10 |
-!- gsomix [~gsomix@83.234.169.131] has quit [Quit: ????? ? ?? ??? (xchat 2.4.5 ??? ??????)] | 00:13 | |
-!- gsomix [~gsomix@83.234.169.131] has joined #shogun | 00:14 | |
-!- emrecelikten [~Anubis@92.44.132.147] has quit [Quit: Leaving.] | 00:15 | |
@sonney2k | blackburn_, 27 | 00:17 |
@sonney2k | nite | 00:17 |
blackburn_ | sonney2k: nite | 00:17 |
blackburn_ | newton svm seems to take a while on my data | 00:17 |
n4nd0 | good night | 00:17 |
blackburn_ | still training | 00:17 |
gsomix | sonney2k, oh, no. all the changes I'll post tomorrow. | 00:21 |
gsomix | i fall asleep =___= | 00:21 |
gsomix | good night, guys | 00:23 |
blackburn_ | nite | 00:23 |
n4nd0 | good night | 00:24 |
-!- blackburn_ [5bde8018@gateway/web/freenode/ip.91.222.128.24] has quit [Quit: Page closed] | 00:51 | |
n4nd0 | good night people | 00:51 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 260 seconds] | 00:57 | |
-!- av3ngr [av3ngr@nat/redhat/x-epzrsrwbdomgshvs] has joined #shogun | 01:32 | |
-!- maximus [86544bc9@gateway/web/freenode/ip.134.84.75.201] has joined #shogun | 02:56 | |
-!- maximus is now known as Guest79494 | 02:56 | |
-!- Guest79494 [86544bc9@gateway/web/freenode/ip.134.84.75.201] has quit [Client Quit] | 02:57 | |
-!- flxb [~cronor@g231230100.adsl.alicedsl.de] has quit [Quit: flxb] | 02:57 | |
-!- mohit [86544bc9@gateway/web/freenode/ip.134.84.75.201] has joined #shogun | 03:05 | |
mohit | Hi | 03:05 |
mohit | is there anybody from shogun here? | 03:05 |
mohit | I have a query regarding gsoc? | 03:07 |
-!- mohit [86544bc9@gateway/web/freenode/ip.134.84.75.201] has quit [Quit: Page closed] | 03:14 | |
-!- pluskid [~chatzilla@li400-235.members.linode.com] has joined #shogun | 03:18 | |
-!- pluskid_ [chatzilla@111.120.23.144] has joined #shogun | 04:46 | |
-!- pluskid [~chatzilla@li400-235.members.linode.com] has quit [Ping timeout: 276 seconds] | 04:50 | |
-!- pluskid_ is now known as pluskid | 04:50 | |
-!- pluskid [chatzilla@111.120.23.144] has quit [Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120312181643]] | 05:39 | |
-!- Johannes_ [55e5c1c3@gateway/web/freenode/ip.85.229.193.195] has joined #shogun | 05:49 | |
Johannes_ | Morning, do I have to install examples seperatly? Been trying all morning to run one :) | 05:59 |
Johannes_ | Or is it perhaps as simple as that it's only included in the source code version, not from the package. | 06:01 |
Johannes_ | yup, was as simple as that. | 06:07 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 07:05 | |
-!- Afan [~Afan@c-76-29-50-209.hsd1.il.comcast.net] has joined #shogun | 07:22 | |
@sonney2k | Johannes_, use this to get them from git git submodule update --init | 08:08 |
Johannes_ | Thank you, downloaded the source code instead, might still try that though. | 08:09 |
Johannes_ | still can't run the java-modular examples though, probably need to install something additional because I couldn't locate modshogun.jar | 08:11 |
@sonney2k | Johannes_, did it compile the modshogun.jar | 08:16 |
@sonney2k | ? | 08:16 |
@sonney2k | if so it should be somewhere under shogun/java_modular | 08:16 |
@sonney2k | and you need to have CLASSPATH set ... | 08:16 |
@sonney2k | morning n4nd0 | 08:17 |
Johannes_ | not sure if it compiled the modshogun.jar, used the package, wasn't so specific at install. | 08:23 |
-!- flxb [~cronor@g231228114.adsl.alicedsl.de] has joined #shogun | 08:23 | |
@sonney2k | use find ./ -name '*.jar' to figure out if it did | 08:25 |
@sonney2k | gtg | 08:25 |
@sonney2k | l8r | 08:25 |
Johannes_ | yea tried that already, doesen't look like it installed it | 08:29 |
Johannes_ | thanks for the help. | 08:29 |
gsomix | sonney2k, hi | 08:31 |
n4nd0 | sonney2k: good morning | 08:42 |
-!- Afan [~Afan@c-76-29-50-209.hsd1.il.comcast.net] has quit [Ping timeout: 246 seconds] | 08:46 | |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has quit [Ping timeout: 276 seconds] | 09:05 | |
-!- av3ngr [av3ngr@nat/redhat/x-epzrsrwbdomgshvs] has quit [Quit: That's all folks!] | 09:08 | |
CIA-64 | shogun: pluskid master * r580d1f8 / (3 files): Use the built-in get_unique_labels instead of wrting a new function. - http://git.io/6whL1A | 09:23 |
CIA-64 | shogun: pluskid master * r49b6e2e / (5 files): Refine the code according to coding convention. - http://git.io/A1opPw | 09:23 |
CIA-64 | shogun: Soeren Sonnenburg master * r874ebc5 / (9 files in 3 dirs): | 09:23 |
CIA-64 | shogun: Merge pull request #420 from pluskid/normalized-mutual-info | 09:23 |
CIA-64 | shogun: Normalized mutual information - http://git.io/XYUlvA | 09:23 |
CIA-64 | shogun: pluskid master * r57d869d / (6 files in 3 dirs): Added normalized mutual information evaluation for clustering. - http://git.io/nLyPtg | 09:23 |
CIA-64 | shogun: pluskid master * r9473a5d / examples/undocumented/python_modular/evaluation_clustering.py : Remove debug code. - http://git.io/dqoG3w | 09:23 |
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun | 09:24 | |
sonne|work | gsomix: one further idea - you could replace/remove the current CSet implementation with your hash set one | 09:32 |
sonne|work | gsomix: and even one more: in shogun we have SGArray/SGArray2 etc - these all can be removed and replaced by SGVector/SGMatrix/SGNDArray | 09:33 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 09:35 | |
-!- gsomix [~gsomix@83.234.169.131] has quit [Ping timeout: 260 seconds] | 09:36 | |
n4nd0 | sonne|work: one doubt about the design related to covertree in KNN | 09:43 |
n4nd0 | sonne|work: it would be really nice if CDistance would support queries for distances between vectors in the same features object | 09:44 |
n4nd0 | sonne|work: now we can do distance(a, b) that gives the distance between vector a in lhs and vector b in rhs | 09:44 |
sonne|work | what is this needed for? | 09:45 |
n4nd0 | sonne|work: what I mean is that it would be useful if we can do distance_lhs(a, b) | 09:45 |
n4nd0 | so in KNN | 09:45 |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 09:45 | |
n4nd0 | we initilize the distance with train vectors | 09:45 |
sonne|work | you need it for covertrees + KNN? | 09:45 |
n4nd0 | so we have distance(train, train) | 09:46 |
n4nd0 | yes | 09:46 |
n4nd0 | but if we later apply we will have distance(train, test) | 09:46 |
n4nd0 | so if for any reason train is called again, it will fail | 09:46 |
sonne|work | yeah but you can do that in the train() function | 09:46 |
sonne|work | there you have distance(train,train) | 09:47 |
n4nd0 | aham ok | 09:47 |
n4nd0 | and when it finishes, should it come back to the initial state | 09:47 |
n4nd0 | e.g. if train is called after apply, then at exit it should be again distance(train, test) | 09:48 |
n4nd0 | or is it not necessary? | 09:48 |
sonne|work | but apart from that I don't mind any distance_lhs() / distance_rhs or kernel_lhs/rhs functions | 09:48 |
sonne|work | not necessary | 09:48 |
sonne|work | it usually is KNN.train() | 09:48 |
sonne|work | then only KNN.test | 09:48 |
sonne|work | apply I mean | 09:49 |
n4nd0 | well I say it because if is done something like | 09:50 |
n4nd0 | KNN.train(train1) | 09:50 |
n4nd0 | KNN.apply(test1) | 09:50 |
n4nd0 | KNN.train(train2) | 09:50 |
n4nd0 | KNN.apply() | 09:50 |
n4nd0 | then it would make the second apply taking train2 as the vectors to classify | 09:51 |
n4nd0 | gtg now, bye! | 09:54 |
sonne|work | n4nd0: yes exactly that is how it is supposed to be | 10:06 |
n4nd0 | sonne|work: ok no problem then, no need to have this distance_lhs and distance_rhs I talked about | 10:19 |
sonne|work | of course if this behavior is not logical - we have to change things | 10:20 |
sonne|work | it has been so since the dawn of days so I might be too blind already to see the obvious | 10:20 |
-!- uricamic [9320543b@gateway/web/freenode/ip.147.32.84.59] has joined #shogun | 10:42 | |
sonne|work | hi uricamic | 11:10 |
uricamic | hi | 11:10 |
n4nd0 | well guys I am going to travel now | 11:18 |
n4nd0 | talk to you later! | 11:19 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 11:19 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 11:31 | |
blackburn | wazzup | 11:31 |
sonne|work | blackburn: what was up with newtonsvm? | 11:52 |
blackburn | sonne|work: crashed with not enough memory actually, I should try different data | 11:52 |
sonne|work | hmmhh memory leak or other bug? | 11:52 |
-!- PhilTillet [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 11:53 | |
blackburn | sonne|work: no, can be caused by lack of memory actually | 11:54 |
blackburn | the data is rather big | 11:54 |
blackburn | I'll ceck again | 11:54 |
blackburn | check* :) | 11:58 |
-!- PhilTillet [~Philippe@tillet-p42154.maisel.int-evry.fr] has quit [Ping timeout: 276 seconds] | 11:58 | |
-!- pythonroar_ [d2198538@gateway/web/freenode/ip.210.25.133.56] has joined #shogun | 12:06 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 12:23 | |
-!- Varagrawal [0e8b7952@gateway/web/freenode/ip.14.139.121.82] has joined #shogun | 13:12 | |
Varagrawal | Hey guys | 13:13 |
Varagrawal | Sorry I wasn't active here for the past few days. Had my GRE this last week | 13:13 |
Varagrawal | I am interested in helping out for the Fast Newton Based SVM solver and the ECOC porting projects | 13:17 |
Varagrawal | I am quite new to statistical machine learning, but I have some basic to intermediate experience with it as well as I am quite familiar with Matlab and Octave | 13:18 |
Varagrawal | Is Soeren sir around? | 13:18 |
blackburn | hi | 13:19 |
blackburn | newton based svm has been recently merged | 13:19 |
blackburn | if you want to work on ECOC you'd have to hurry up, gsoc proposal deadline is tomorrow | 13:25 |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has quit [Ping timeout: 244 seconds] | 13:41 | |
Varagrawal | recently merged? | 13:50 |
Varagrawal | You mean it's been done already?? | 13:50 |
Varagrawal | :( | 13:50 |
-!- shogun-toolbox [~shogun@7nn.de] has quit [Ping timeout: 245 seconds] | 13:50 | |
--- Log closed Thu Apr 05 13:50:34 2012 | ||
--- Log opened Thu Apr 05 13:50:42 2012 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 13:50 | |
-!- Irssi: #shogun: Total of 13 nicks [1 ops, 0 halfops, 0 voices, 12 normal] | 13:50 | |
-!- Irssi: Join to #shogun was synced in 7 secs | 13:50 | |
Varagrawal | anybody out there? | 13:55 |
blackburn | Varagrawal: yes, newton svm is done | 13:58 |
Varagrawal | :( | 13:59 |
Varagrawal | cool, anything else that might seem a good fit for me? | 13:59 |
blackburn | what do you prefer? | 14:00 |
blackburn | we always welcome new examples | 14:01 |
Varagrawal | well I am not very experienced in ML as I am self-taught and still an undergrad | 14:02 |
Varagrawal | but when Shogun was selected for GSoC, this was the only project I had my eye one. | 14:02 |
-!- gsomix [~gsomix@188.168.14.72] has joined #shogun | 14:03 | |
gsomix | hi. | 14:03 |
blackburn | hey there | 14:04 |
gsomix | today's lecture about philosophy was interesting. | 14:07 |
blackburn | Varagrawal: strange you have chosen shogun then :) | 14:08 |
Varagrawal | I am interested in Machine Learning and want to become as good as you guys | 14:08 |
Varagrawal | I just got killed coz I had my midsem exams in mid-march and then my GRE the previous week :'( | 14:09 |
blackburn | I still think to develop some nice graphical example is a good way to learn | 14:09 |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 14:11 | |
sonne|work | gsomix: seen the 2 other ideas I had? | 14:11 |
Varagrawal | I am going through the tutorial pages of shogun | 14:11 |
Varagrawal | I had cloned the repo | 14:11 |
Varagrawal | but then I ws getting this weird error for one application. Nando know about it and he had absolutely no clue | 14:12 |
gsomix | sonne|work, CSet -> hash set? | 14:12 |
blackburn | Varagrawal: we are on the way to fix applications, you should rather check examples | 14:12 |
blackburn | applications are pretty complex and domain specific | 14:12 |
Varagrawal | I was just running them to check how breaking some code would affect the application | 14:13 |
sonne|work | gsomix: yes and SGArray removal | 14:13 |
sonne|work | (replace by SGVector) | 14:13 |
sonne|work | gsomix: but now I really think this is enough... | 14:14 |
sonne|work | except if you intend to work 60+ hrs :D | 14:14 |
blackburn | sonne|work: he will be able! | 14:15 |
blackburn | sonne|work: 90 is ok for him | 14:15 |
gsomix | sonne|work, i dont have 60+ hrs. :) i think such changes can be done if there is free time. | 14:15 |
gsomix | blackburn, >:3 | 14:16 |
blackburn | gsomix: you do! I recall you wanted to work for a month w/o sleeping | 14:16 |
sonne|work | I am sure that with the help of vodka you both will manage | 14:17 |
gsomix | blackburn, okay, master. =___= | 14:17 |
blackburn | sonne|work: the vodka | 14:17 |
sonne|work | yes vodking? | 14:17 |
sonne|work | ha :D | 14:17 |
blackburn | vod-king? | 14:18 |
blackburn | gsomix: I have some small task for you | 14:18 |
blackburn | all you need is to implement distributed linear svm core faster than liblinear | 14:19 |
blackburn | should be easy | 14:19 |
gsomix | sonne|work, anyway I want to work on shogun after GSoC. It's approximately infinity time. :] | 14:19 |
gsomix | blackburn, wut? you're at work? | 14:20 |
blackburn | gsomix: I am | 14:20 |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has joined #shogun | 14:23 | |
gsomix | professor of philosophy said that the Russian and German languages are better suited for metaphysics than others | 14:23 |
gsomix | fun fact | 14:24 |
blackburn | I guess the reason is he knows only these two ones | 14:24 |
gsomix | blackburn, and a little Sanskrit and English | 14:26 |
blackburn | really? he tends to pronounce english words in a funny way | 14:27 |
blackburn | however this fact could arise for real | 14:27 |
blackburn | I mean native language determines the way of thinking often | 14:28 |
gsomix | I see | 14:29 |
blackburn | gsomix: and even programming languages do the same | 14:29 |
-!- sidsniper [~piggy@14.139.82.6] has joined #shogun | 14:38 | |
PhilTillet | Hello everybody | 14:39 |
blackburn | hi | 14:39 |
PhilTillet | what's up? | 14:39 |
sidsniper | sky :P | 14:39 |
PhilTillet | lol. | 14:40 |
PhilTillet | and avian carriers? | 14:40 |
PhilTillet | :D | 14:40 |
sidsniper | i was wondering how many slots will shogun get this time? | 14:40 |
sonne|work | we are too | 14:40 |
blackburn | hah | 14:40 |
PhilTillet | rand() slots | 14:43 |
-!- sidsniper [~piggy@14.139.82.6] has left #shogun [] | 14:43 | |
-!- sidsnipe [~piggy@14.139.82.6] has joined #shogun | 14:43 | |
blackburn | no, I don't like rand stuff there | 14:45 |
-!- muddo [~muddo@gateway/tor-sasl/muddo] has joined #shogun | 14:46 | |
PhilTillet | :D let's just hope it will be a Dirac Distribution | 14:46 |
PhilTillet | :D | 14:46 |
PhilTillet | rand over dirac distribution is cool | 14:46 |
-!- siddharthk [~siddharth@14.139.82.6] has joined #shogun | 14:47 | |
gsomix | number of slots is determined by position of the planets. | 14:47 |
siddharthk | err is it too late to apply now? | 14:47 |
-!- sidsnipe [~piggy@14.139.82.6] has left #shogun [] | 14:48 | |
blackburn | siddharthk: due date is tomorrow | 14:48 |
siddharthk | are there any free projects for which i can apply for? | 14:50 |
blackburn | siddharthk: what do you mean free? ;) | 14:51 |
siddharthk | which has less competition :P | 14:51 |
PhilTillet | sonney2k, maybe I can release a first iteration of the OpenCL patch tomorrow, but design will really not be optimal :D As computing kernel values one per one is completely inefficient (and the whole shogun design is kinda based on that :D) | 14:51 |
siddharthk | previous year me n u applied for same project :P | 14:51 |
siddharthk | i got axed :/ | 14:52 |
blackburn | siddharthk: yeah | 14:52 |
PhilTillet | oops, should have been sonne|work rather than sonney2k maybe :D | 14:52 |
blackburn | siddharthk: ok libqp and multiclass probably have less competition there | 14:53 |
siddharthk | i can see that there are more people now :) | 14:53 |
blackburn | however I can't be sure | 14:53 |
siddharthk | ok let me check ideas page | 14:53 |
blackburn | not really last year there were 70+ appliants | 14:53 |
siddharthk | i mean on irc | 14:53 |
blackburn | ah | 14:54 |
blackburn | yeah may be | 14:54 |
siddharthk | i hope patch will not be a problem this time | 14:55 |
-!- pythonroar_ [d2198538@gateway/web/freenode/ip.210.25.133.56] has quit [Ping timeout: 245 seconds] | 14:55 | |
blackburn | what do you mean? | 14:55 |
siddharthk | i mean previous year patches would count right? :P | 14:57 |
blackburn | yes however I think to have this year patches would be better ;) | 14:59 |
harshit_ | blackburn: Can i send a pull request for shougun-toolbox/shogun-data ? | 15:01 |
blackburn | harshit_: yeah probably but you would need to fork it too | 15:01 |
blackburn | what do you need to change? | 15:01 |
harshit_ | I tested my octave_modular example on splice dataset, So can i add splice dataset in it | 15:02 |
harshit_ | its not too big | 15:02 |
harshit_ | and for now I have changed my octave example on toydataset | 15:02 |
blackburn | that's ok | 15:02 |
blackburn | we use toy data in examples | 15:02 |
harshit_ | okay .. | 15:03 |
PhilTillet | blackburn, you once told me about Dimension Reduction on OpenCL but I was not here :D | 15:04 |
blackburn | PhilTillet: yes that is something of interest probably | 15:04 |
PhilTillet | but, this is only simple BLAS Matrix operations right? | 15:04 |
PhilTillet | I mean, for PCA at least | 15:04 |
blackburn | yes | 15:04 |
PhilTillet | I think it will be very easy to port in a few months when ViennaCL will be more advanced :p they plan to include some eigenvalues algorithms on GPU | 15:05 |
PhilTillet | then it would be possible without writing any OpenCL code :p | 15:05 |
-!- harshit_ [~harshit@182.68.94.221] has quit [Quit: Leaving] | 15:05 | |
blackburn | I don't think they will include efficient sparse eigensolver | 15:06 |
PhilTillet | hmmm | 15:06 |
PhilTillet | this was a GSoC project last year as far as I remember | 15:06 |
blackburn | LLE/LTSA/HLLE/.... are sparse | 15:06 |
blackburn | really? I'll take a look later | 15:06 |
PhilTillet | hmmm yes indeed | 15:07 |
PhilTillet | maybe it was not sparse | 15:07 |
PhilTillet | It was mainly Lanczos and Arnoldi's | 15:07 |
blackburn | lanczos and arnoldi stuff makes use of matrix-vector product | 15:08 |
blackburn | so they could be sparse | 15:08 |
PhilTillet | yep | 15:08 |
PhilTillet | there is sparse matrix-vector implemented | 15:08 |
blackburn | however in shogun I had to integrate superlu | 15:08 |
blackburn | there was need of sparse solver | 15:08 |
PhilTillet | well | 15:08 |
PhilTillet | there is GPU LU decomposition on ViennaCL too I think | 15:09 |
PhilTillet | :p | 15:09 |
blackburn | superlu is sparse (again this word) ;) | 15:09 |
PhilTillet | haha | 15:09 |
PhilTillet | sparsity yay | 15:09 |
PhilTillet | actually i'm not an expert concerning ViennaCL, but if you have some question concerning what linalg operation is implemented, I can relay them to the project leader | 15:10 |
blackburn | ok thanks | 15:11 |
blackburn | okay gtg | 15:11 |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed] | 15:11 | |
-!- gsomix [~gsomix@188.168.14.72] has quit [Ping timeout: 260 seconds] | 15:12 | |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 15:26 | |
CIA-64 | shogun: Harshit Syal master * r9241292 / (2 files in 2 dirs): Added brief intro to NewtonSVM and added octave_modular example - http://git.io/xpkZDQ | 15:36 |
CIA-64 | shogun: Harshit Syal master * rfc37445 / src/shogun/classifier/svm/NewtonSVM.h : Removed some typo - http://git.io/QQrc3g | 15:36 |
CIA-64 | shogun: Soeren Sonnenburg master * r39e947f / (2 files in 2 dirs): | 15:36 |
CIA-64 | shogun: Merge pull request #421 from harshitsyal/master | 15:36 |
CIA-64 | shogun: Newton SVM : improve documentation - http://git.io/wJaZLg | 15:36 |
-!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 260 seconds] | 15:40 | |
-!- ani1 [dcff023b@gateway/web/freenode/ip.220.255.2.59] has joined #shogun | 15:43 | |
-!- ani1 [dcff023b@gateway/web/freenode/ip.220.255.2.59] has quit [Ping timeout: 245 seconds] | 16:05 | |
-!- gsomix [~gsomix@188.168.3.37] has joined #shogun | 16:07 | |
-!- Kshiteesh [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has joined #shogun | 16:32 | |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 16:32 | |
-!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 16:43 | |
-!- PhilTillet [~Philippe@npasserelle10.minet.net] has quit [Ping timeout: 246 seconds] | 16:44 | |
-!- sonne|work [~sonnenbu@194.78.35.195] has quit [Quit: Leaving.] | 16:49 | |
PhilTillet_ | sonney2k, I've got some lil' problem with the ./configure | 16:58 |
PhilTillet_ | when I used ./consigure --cxxflags=-DUSE_OPENCL, it compiles fine but removes the -fPIC | 16:59 |
PhilTillet_ | so it fails at linking | 16:59 |
PhilTillet_ | (but at least it finally compiles, but cannot test it :D) | 17:02 |
-!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 264 seconds] | 17:09 | |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 17:14 | |
-!- Kshiteesh [~KMH@c-24-118-111-180.hsd1.mn.comcast.net] has quit [Ping timeout: 265 seconds] | 17:20 | |
@sonney2k | PhilTillet_, well then use --cxxflags="-DUSE_OPENCL -fPIC" | 17:20 |
PhilTillet_ | oh of course | 17:21 |
PhilTillet_ | i had forgotten the quotes... kinda tired :D | 17:21 |
PhilTillet_ | ./configure --cxxflags="-DUSE_OPENCL -fPIC" => Error: Unknown parameter: -fPIC | 17:22 |
PhilTillet_ | tricky | 17:22 |
PhilTillet_ | Sounds like the configure script is trolling me. | 17:23 |
@sonney2k | PhilTillet_, argh then use CXXFLAGS=-DUSE_OPENCL ./configure | 17:26 |
-!- n4nd0 [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 17:38 | |
-!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 265 seconds] | 17:46 | |
-!- flxb [~cronor@g231228114.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 18:05 | |
-!- n4nd0 [~androirc@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Remote host closed the connection] | 18:07 | |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 18:13 | |
-!- flxb [~cronor@g231228114.adsl.alicedsl.de] has joined #shogun | 18:20 | |
shogun-buildbot | build #438 of ruby_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/438 blamelist: sonne@debian.org, harshit.syal@nsitonline.in | 18:30 |
-!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 264 seconds] | 18:35 | |
PhilTillet_ | sonney2k, how can I add files to the build? | 18:37 |
-!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has quit [Read error: Operation timed out] | 19:00 | |
-!- PhilTillet_ [~Philippe@npasserelle10.minet.net] has joined #shogun | 19:16 | |
-!- PhilTillet_ [~Philippe@npasserelle10.minet.net] has quit [Ping timeout: 244 seconds] | 19:23 | |
-!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has joined #shogun | 19:34 | |
-!- PhilTillet_ [~Philippe@tillet-p42154.maisel.int-evry.fr] has joined #shogun | 19:36 | |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 19:39 | |
-!- Kshiteesh [~KMH@nat-portal-160-94-47-16.uofm.wireless.umn.edu] has joined #shogun | 19:40 | |
-!- uricamic [9320543b@gateway/web/freenode/ip.147.32.84.59] has quit [Quit: Page closed] | 19:40 | |
wiking | PhilTillet_: well if it's a cpp file it'll just compile it | 19:50 |
wiking | but of course if you want to add it to the swig interface as well | 19:51 |
wiking | then you'll have to edit some files for includes.. | 19:51 |
-!- harshit_ [~harshit@182.68.94.221] has quit [Ping timeout: 260 seconds] | 19:57 | |
PhilTillet_ | the swig interface ? :p | 20:00 |
PhilTillet_ | isn't autotools used? | 20:00 |
wiking | well if you want to support modules | 20:12 |
wiking | like python_modular | 20:12 |
wiking | and such then you'll want swig | 20:12 |
wiking | and for that you have to do some extra work... | 20:12 |
PhilTillet_ | yes I know a bit of swig, headache | 20:13 |
@sonney2k | no autotools all magic | 20:16 |
PhilTillet_ | sounds like fun | 20:18 |
PhilTillet_ | XD | 20:18 |
PhilTillet_ | oh, and some other question, I used #ifdef USE_OPENCL about everywhere, but then once you have the library you have to do #define USE_OPENCL so that you see the functions, any way to avoid this? :p | 20:19 |
@sonney2k | PhilTillet_, that can be taken care of later via some configure option | 20:25 |
PhilTillet_ | oh yes, true | 20:26 |
PhilTillet_ | I hate coding with #ifdef | 20:26 |
PhilTillet_ | cause usually ide don't parse the #ifdef'ed code | 20:27 |
PhilTillet_ | >< | 20:27 |
-!- harshit_ [~harshit@182.68.94.221] has joined #shogun | 20:27 | |
n4nd0 | PhilTillet_: what IDE do you use? | 20:27 |
@sonney2k | PhilTillet_, just to explain: in shogun the kernel matrix is assumed to *not* fit in memory | 20:27 |
PhilTillet_ | sonney2k, I know, but I cannot implement that in 1 week :D Making it work for big dataset is the whole point of the SoC | 20:28 |
PhilTillet_ | :p | 20:28 |
@sonney2k | that is why you need to call kernel(i,j) most of the time. actually even examples must not be in memory | 20:28 |
@sonney2k | and are computed on-the-fly | 20:28 |
PhilTillet_ | n4nd0, KDevelop | 20:28 |
@sonney2k | ok | 20:29 |
PhilTillet_ | I mean it will require some clever design, but I am sure it is possible to do some queueing system, with flushing | 20:29 |
PhilTillet_ | and, even if it is not, dot product is still faster on GPU than on CPU | 20:29 |
PhilTillet_ | usually | 20:30 |
PhilTillet_ | it should suffer from bad caching though :p | 20:30 |
-!- Johannes_ [55e5c1c3@gateway/web/freenode/ip.85.229.193.195] has quit [Ping timeout: 245 seconds] | 20:32 | |
@sonney2k | well you certainly can do speedups for special cases - like when simplefeatures are used and the feature matrix is in memory | 20:32 |
PhilTillet_ | yes | 20:33 |
PhilTillet_ | and when simplefeatures matrix do not fit in memory, it should be possible to transfer it chunks by chunk, for apply at least | 20:35 |
PhilTillet_ | even though the training might get more complicated | 20:35 |
PhilTillet_ | and, even for training, something clever might be to use hybrid cpu/gpu algorithm | 20:35 |
PhilTillet_ | but synchronization might then be a pain | 20:36 |
@sonney2k | PhilTillet_, well lets focus on testing first | 20:39 |
@sonney2k | training usually has to be done once | 20:39 |
@sonney2k | testing inf times | 20:39 |
PhilTillet_ | true :d | 20:39 |
PhilTillet_ | I mean, for testing, examples are independent, so it might be possible to do something like : "Transfer chunk 1 from CPU to GPU, compute chunk 2 on CPU, get chunk 1 result, transfer chunk 3 to GPU, compute chunk 4 on CPU | 20:41 |
PhilTillet_ | things like that | 20:41 |
PhilTillet_ | i'll spend some time again on my gsoc timeline, to focus on hybrid CPU / GPU algorithm | 20:42 |
harshit_ | sonney2k: Is fast k-Means algo implemented in shogun ? I saw this in issues section of github . | 20:45 |
harshit_ | link:https://github.com/shogun-toolbox/shogun/issues/338 | 20:46 |
@sonney2k | harshit_, no unfortunately not | 20:46 |
harshit_ | sonney2k: Its mentioned by authors that you cannot redistribute this code . but still can we port it ? | 20:47 |
n4nd0 | I am working in cover tree integration now | 20:47 |
@sonney2k | n4nd0, for KNN right? | 20:47 |
n4nd0 | sonney2k: yeah | 20:48 |
@sonney2k | harshit_, well the algoritm / paper is very simple | 20:48 |
@sonney2k | probably devil is in the detail but still one can compare outputs of what they get to what you would implement | 20:48 |
harshit_ | n4nd0 : is cover tree , in anyway thats related to fast k-means .. i think it wound not but still confirming .. | 20:49 |
harshit_ | coz I am always confused about differences in KNN and K means except one is unsupervised | 20:50 |
harshit_ | sonney2k: yeah algo is really simple..got one more thing to work on :) | 20:50 |
@sonney2k | yeah covertree would speed up k-means too - in the end you need distances for that too | 20:56 |
n4nd0 | it is turning out to be a bit cumbersome though :S | 20:57 |
-!- siddharthk [~siddharth@14.139.82.6] has quit [] | 21:20 | |
flxb | Is there a possibility to save a GMNPSVM without using pickle? pickle is too slow, because the classifier is 15GB and cPickle has a memory bug. | 21:40 |
@sonney2k | flxb, look at serialization_complex_example.py | 21:42 |
-!- Kshiteesh [~KMH@nat-portal-160-94-47-16.uofm.wireless.umn.edu] has quit [Ping timeout: 252 seconds] | 21:46 | |
PhilTillet_ | sonney2k, is the kernel matrix always double precision? | 21:57 |
-!- Varagrawal [0e8b7952@gateway/web/freenode/ip.14.139.121.82] has quit [Ping timeout: 245 seconds] | 22:00 | |
flxb | thanks, sonney2k | 22:17 |
-!- blackburn [~qdrgsm@83.234.54.186] has joined #shogun | 22:18 | |
n4nd0 | blackburn: hey! I got a question here | 22:22 |
blackburn | yeah | 22:22 |
n4nd0 | blackburn: I am analizing with valgrind a python example | 22:22 |
gsomix | blackburn, hello, master | 22:22 |
blackburn | gsomix: lol | 22:23 |
gsomix | >:3 | 22:23 |
blackburn | n4nd0: sooo | 22:23 |
n4nd0 | blackburn: http://snipt.org/uhaji0 | 22:23 |
blackburn | aham | 22:24 |
n4nd0 | blackburn: I am giving you the code where it happens, w8 | 22:24 |
n4nd0 | https://github.com/shogun-toolbox/shogun/blob/master/src/shogun/classifier/KNN.cpp | 22:24 |
n4nd0 | line 235 there | 22:25 |
n4nd0 | it appears 296 in valgrind because I have executed it in my local version with CoverTree stuff, but this is not related | 22:26 |
n4nd0 | no CoverTree used there yet | 22:26 |
blackburn | ok pretty strange | 22:26 |
blackburn | I'd say it is a false alarm there | 22:26 |
@sonney2k | n4nd0, please change the line SGMatrix<int32_t>(output,num_lab,m_k) | 22:26 |
@sonney2k | to SGMatrix<int32_t>(output,num_lab,m_k,true) | 22:26 |
@sonney2k | then it should be fixed | 22:26 |
blackburn | yeah had the same idea | 22:27 |
n4nd0 | sonney2k, blackburn: you guys are pretty good :) | 22:28 |
n4nd0 | fixed | 22:28 |
n4nd0 | sonney2k: do you have a good explanation for it? | 22:29 |
@sonney2k | yeah - do valgrind only w/ debugging flags on + no optimizations | 22:30 |
@sonney2k | otherwise line nr might not fit | 22:30 |
blackburn | what is the reproducing example? | 22:30 |
n4nd0 | classifier_knn_python_modular.py | 22:30 |
blackburn | ah | 22:31 |
n4nd0 | sonney2k: I didn't understand ... I mean the line valgrind was right I think | 22:31 |
blackburn | n4nd0: true stands for do_free | 22:32 |
@sonney2k | then I don't understand your question :) | 22:32 |
n4nd0 | sonney2k: because the part of memory not freed is in output = SG_MALLOC(...) | 22:32 |
n4nd0 | sonney2k: yeah the question is, why do we need to set explicitily this do_free? | 22:33 |
PhilTillet_ | sonney2k, the patch works and should be issued soon :p | 22:37 |
PhilTillet_ | it is in a new branch, but do you want it as a single commit? | 22:39 |
@sonney2k | n4nd0, this do_free needs to be set whenever you alloc sth that is returned / if you just pass a ptr then it should be false | 22:39 |
@sonney2k | but this is what I would like to change with the refcounting stuff | 22:39 |
n4nd0 | sonney2k: ok, I understand | 22:44 |
gsomix | sonney2k, there is small pull request. It's related to python3 interface. I hope I will send fixes for examples tomorrow. | 22:45 |
@sonney2k | gsomix, ok | 22:45 |
@sonney2k | PhilTillet_, not sure what you mean - something we can review where each commit compiles would be nice | 22:46 |
PhilTillet_ | ok, my question was dumb indeed | 22:46 |
PhilTillet_ | i'll discard the bugfix commits :D | 22:47 |
@sonney2k | (additional / unrelated bugfix commits as extra pull requests please) | 22:48 |
@sonney2k | n4nd0, please send a pull request with that one-liner KNN patch | 22:48 |
PhilTillet_ | sonney2k, what do you mean? | 22:48 |
@sonney2k | bugfixes you did on the side | 22:48 |
PhilTillet_ | like bugfixes on my generator? | 22:48 |
n4nd0 | sonney2k: do you want me to prepare an example of performance improvement first? | 22:48 |
PhilTillet_ | oops | 22:49 |
PhilTillet_ | not generator | 22:49 |
PhilTillet_ | bugfix on my opencl support? | 22:49 |
PhilTillet_ | i did no additional shogun bugfix in this opencl branch if this is the question | 22:50 |
PhilTillet_ | :p | 22:50 |
@sonney2k | n4nd0, it would be interesting to know how much the covertree helps or? | 22:50 |
n4nd0 | sonney2k: I was planning to do that and later issue the pull request | 22:51 |
n4nd0 | sonney2k: but I can open it before the example if you want so we can start fixing things I may have done wrong | 22:51 |
@sonney2k | n4nd0, why not | 22:58 |
gsomix | sonney2k, I have some problems with my "const char*" typemap. http://pastebin.com/tiq44rbE | 23:03 |
gsomix | who should take care of freeing memory here? | 23:04 |
gsomix | there is part in python3 doc: http://docs.python.org/py3k/c-api/bytes.html?highlight=pybytes_asstring#PyBytes_AsString | 23:05 |
@sonney2k | gsomix, i don't think const char* typemaps can work | 23:05 |
@sonney2k | but I am not an expert in this either | 23:06 |
@sonney2k | but IMHO char* is automagically swig wrapped | 23:06 |
@sonney2k | so anything doing char* foo() | 23:06 |
@sonney2k | or void foo(char* x) should work | 23:06 |
gsomix | sonney2k, i had problems in modelselection example. | 23:07 |
gsomix | for example, line: param_power_kernel_degree=ModelSelectionParameters("degree") | 23:07 |
gsomix | SWIG incorrectly converted "degree" into const char* | 23:08 |
gsomix | in python3 | 23:08 |
gsomix | then I wrote simple typemap. | 23:09 |
gsomix | at now it works. but I don't know how well. | 23:10 |
@sonney2k | I think we have to remove the const* stuff somewhere then | 23:11 |
n4nd0 | well guys, good night | 23:12 |
n4nd0 | sonney2k: I will issue a PR with covertree stuff tomorrow morning probably | 23:12 |
gsomix | n4nd0, good night | 23:13 |
@sonney2k | n4nd0, nite nite | 23:13 |
-!- n4nd0 [~nando@h121n3-ksg-a11.ias.bredband.telia.com] has quit [Quit: leaving] | 23:13 | |
gsomix | sonney2k, ok. while I will try to research this issue in more details. | 23:14 |
CIA-64 | shogun: Evgeniy Andreev master * r6ea66a8 / src/interfaces/modular/SGBase.i : fixed pickle serialization - http://git.io/c1j6-w | 23:18 |
CIA-64 | shogun: Evgeniy Andreev master * r2320875 / src/interfaces/python_modular/swig_typemaps.i : unicode -> const char* and fixed typemaps for int32 support - http://git.io/9K0bSQ | 23:18 |
CIA-64 | shogun: Soeren Sonnenburg master * r1eae5da / (2 files in 2 dirs): | 23:18 |
CIA-64 | shogun: Merge pull request #423 from gsomix/python3_interface | 23:19 |
CIA-64 | shogun: Updates on python3 interface - http://git.io/PIfgGQ | 23:19 |
gsomix | oh, tomorrow classes will be about programming under linux. | 23:19 |
gsomix | where my physics now? :) | 23:19 |
gsomix | good night, guys | 23:20 |
PhilTillet_ | good night | 23:20 |
@sonney2k | cu | 23:20 |
PhilTillet_ | sonney2k, where can I add the example for using the OpenCL apply() functionnality? | 23:25 |
PhilTillet_ | which directory | 23:25 |
@sonney2k | alright sleep time for me | 23:25 |
PhilTillet_ | okay, good night :p | 23:25 |
@sonney2k | PhilTillet_, which $LANG? | 23:25 |
PhilTillet_ | C++ | 23:25 |
@sonney2k | examples/undocumented/libshogun then | 23:25 |
PhilTillet_ | thanks | 23:25 |
@sonney2k | nite | 23:26 |
PhilTillet_ | night | 23:26 |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 23:27 | |
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 23:42 | |
--- Log closed Fri Apr 06 00:00:19 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!