--- Log opened Wed Oct 31 00:00:17 2012 | ||
@sonney2k | wiking, what's that? | 00:05 |
---|---|---|
@sonney2k | wiking, where is you gcc faq? | 00:06 |
wiking | nowhere yet | 00:21 |
wiking | i'm still looking into how to redesign the faq page | 00:21 |
wiking | ah ok so we dont have sitemap | 00:21 |
wiking | i'll generate one then | 00:21 |
wiking | http://support.google.com/webmasters/bin/answer.py?hl=en&answer=156184 | 00:22 |
@sonney2k | wiking, we might be able to do that automagically based on django urls | 00:25 |
@sonney2k | wiking, it is a bit of work but possible https://docs.djangoproject.com/en/dev/ref/contrib/sitemaps/ | 00:28 |
@sonney2k | ahh what is badly missing from our website is the feature matrix | 00:31 |
@sonney2k | this one here http://shogun-toolbox.org/jmlr10/ | 00:32 |
@sonney2k | it is created based on some google spredsheet | 00:32 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 00:51 | |
-!- ptizoom [~christian@85.210.80.20] has quit [Quit: Ex-Chat] | 01:09 | |
-!- ptizoom_ [~christian@85.210.80.20] has joined #shogun | 01:09 | |
blackburn | IT WORKS | 01:47 |
blackburn | YEAH MOTHERFUCKER | 01:47 |
blackburn | wiking: 1.63s to solve 15k x 15k eigenproblem | 01:51 |
blackburn | that means when I get covertree for nearest neighbor queries I can get to quasilinear embedding cost like 5s for enormous number of vectors | 01:52 |
blackburn | stability issues though :D | 01:53 |
-!- ptizoom_ is now known as ptizoom | 07:55 | |
-!- blackburn [~blackburn@188.168.4.75] has quit [Quit: Leaving.] | 08:15 | |
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun | 08:45 | |
ptizoom | sonney*: shall /perl directory (from shogun root) be used for building and exporting a perl-shogun package (to CPAN)? | 10:10 |
blackburn | sonne|work: let me cite you a VW code line | 10:29 |
blackburn | for (vector<string>::iterator i = all.pairs.begin(); i != all.pairs.end();i++) { | 10:29 |
blackburn | :D | 10:29 |
sonne|work | ptizoom: shogun/src or? | 10:33 |
blackburn | sonne|work: I want to set VW up to date | 10:37 |
blackburn | sonne|work: do you think I should discuss with john? | 10:37 |
sonne|work | he would probably be happy about that | 10:38 |
sonne|work | but he is NY | 10:38 |
blackburn | sonne|work: well hurricane will end someday | 10:38 |
blackburn | :D | 10:38 |
sonne|work | not yet awake I mean | 10:38 |
blackburn | ah | 10:39 |
blackburn | well I use email usually | 10:39 |
blackburn | so you say he would be happy? | 10:39 |
wiking | aaa | 10:39 |
blackburn | I am not really in what you have done with shashwat | 10:40 |
blackburn | it'd be really cool if we just depend on vw | 10:40 |
sonne|work | longer project then | 10:40 |
blackburn | sonne|work: is there any reason to really include the codE? | 10:41 |
blackburn | may be I could help them to create an API I don't know | 10:41 |
blackburn | sonne|work: they have they own development line - so hard to track it | 10:42 |
blackburn | liblinear is not being really developed anymore, vw is | 10:42 |
sonne|work | no feel free | 10:44 |
sonne|work | but lot of work | 10:44 |
blackburn | sonne|work: okay | 10:45 |
blackburn | will write a mail later | 10:45 |
blackburn | john's APIs are pretty cumbersome for me though | 10:46 |
blackburn | I remembed all these struggles with the covertree | 10:46 |
blackburn | I never managed to get it working, only fernando did :D | 10:46 |
sonne|work | well then maybe better choose some other fun project :D | 10:47 |
blackburn | sonne|work: well I got interested | 10:48 |
blackburn | sonne|work: btw I kinda developed a new method for eigendecomposition | 10:48 |
blackburn | sonne|work: you've seen that probably in last night log | 10:48 |
blackburn | it was working already but it took 3 days to realize that kernel distance is k(x,x) + k(y,y) - 2 k(x,y) | 10:49 |
blackburn | and not k(x,y) | 10:49 |
blackburn | :D | 10:49 |
blackburn | so we will get some fun stuff soon | 10:50 |
blackburn | both in standalone and shogun versions | 10:50 |
blackburn | sonne|work: can you stand 3 headers with high std and eigen usage? | 10:51 |
blackburn | flexibility was one of my intentions so I have to live with it | 10:51 |
blackburn | eigen bloats code a little btw | 10:52 |
blackburn | but it is like you have intel MKL with no intel MKL | 10:52 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 11:54 | |
ptizoom | sonney*: from root...together with data,doc,examples,src, application... then perl! unless you find a good place to stuff a Makefile.PL and a t directory? | 13:15 |
sonne|work | blackburn: can't you avoid using eigen/std in headers? | 13:26 |
sonne|work | ptizoom: well I guess at some point we should just move the src/Makefile and configure and everything one layer up but for now | 13:27 |
sonne|work | in src | 13:28 |
ptizoom | sonne*: okey dookey, I build a /src/perl ; typically I guess one put directories like src, perl, python, swig, debian etc... in the root. | 13:32 |
n4nd0 | can someone tell me why #include <eigen3/Eigen/...> doesn't work? | 13:37 |
n4nd0 | I have /usr/local/include/eigen3 | 13:37 |
n4nd0 | and also /usr/include/eigen3 although I don't really know why both of them... | 13:38 |
n4nd0 | I made a symbolic link of /usr/local/include/eigen3/Eigen in /usr/local/include and then I got to compile with the directive #include <Eigen/...> | 13:38 |
n4nd0 | but still, I do not get why #include <eigen3/Eigen/...> fails | 13:38 |
sonne|work | n4nd0: maybe eigen3 is including sth with <Eigen/xxx> too? | 13:48 |
sonne|work | ptizoom: if it doesn't make sense we will change it later | 13:48 |
n4nd0 | sonne|work: what do you mean? | 13:48 |
n4nd0 | eigen3 is directory in /usr/local/include that includes another directory called Eigen | 13:49 |
sonne|work | yeah but /usr/local/include/eigen3 is not in the include path | 14:36 |
sonne|work | so when you include <eigen3/Eigen/sth> | 14:36 |
sonne|work | and sth includes <Eigen/whateve> it cannot find whatever | 14:36 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 14:47 | |
wiking | n4nd0 | 17:17 |
wiking | ah shit | 17:17 |
wiking | but yeah he is doing it wrong | 17:17 |
wiking | he should include -issytem /user/local/include/eigen3 | 17:17 |
wiking | and then #include <Eigen... | 17:17 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 17:50 | |
-!- blackburn_ [d5578acb@gateway/web/freenode/ip.213.87.138.203] has joined #shogun | 17:51 | |
blackburn_ | sonne|work: I can hide them to cpp adapter | 17:52 |
blackburn_ | sonne|work: but there are templates | 17:52 |
blackburn_ | not specialized so can't put realization anywhere | 17:53 |
blackburn_ | sonne|work: that's all negotiable though but I still like functors more than crazy function pointers | 17:55 |
blackburn_ | so not really sure I can avoid templating | 17:55 |
blackburn_ | sonne|work: what I am sure is that all libraries should use reverse interface | 17:59 |
-!- fmailhot [~fmailhot@c-24-10-116-3.hsd1.ca.comcast.net] has joined #shogun | 18:01 | |
fmailhot | Hi all...what's the appropriate Python interface for multiclass logistic regression trained via SGD? (assuming Shogun has this) I'd like to do some Shogun/scikit-learn comparisons... | 18:09 |
blackburn_ | fmailhot: I have bad news - your assumption is a little wrong :) | 18:13 |
blackburn_ | fmailhot:but modular is the only interface one should use | 18:13 |
fmailhot | OK...python_modular++...got it. | 18:14 |
fmailhot | And...multiclass SGD-trained LR? | 18:14 |
blackburn_ | fmailhot: you need exactly that method? | 18:15 |
blackburn_ | fmailhot: I can suggest you to try MulticlassLibLinear | 18:16 |
blackburn_ | L2 regularized SVM Crammer-Singer that is | 18:16 |
blackburn_ | swap(SVM,Crammer-Singer) | 18:16 |
fmailhot | I'm hoping for something that can be trained online, in a streaming data kind of scenario... | 18:18 |
blackburn_ | fmailhot: LaRank | 18:18 |
fmailhot | Oh sorry, and most to the point, I'd like well-calibrated probabilities as outputs (hence my initial request for LR) | 18:19 |
blackburn_ | fmailhot: ah I see | 18:19 |
blackburn_ | fmailhot: if you can still live with batch mode I have one more suggestion | 18:20 |
* fmailhot is all ears | 18:20 | |
blackburn_ | a minute | 18:21 |
blackburn_ | forgot the name of class I wrote :D | 18:21 |
blackburn_ | fmailhot: https://github.com/shogun-toolbox/shogun/blob/master/src/shogun/multiclass/MulticlassTreeGuidedLogisticRegression.h | 18:21 |
blackburn_ | fmailhot: it is tree group lasso multiclass logistic regression | 18:22 |
blackburn_ | yeah it is funny we have such complex algorithms and have no SGD | 18:22 |
blackburn_ | :D | 18:22 |
blackburn_ | fmailhot: I think I can implement basic multiclass logistic regression in next 2-3 days if it is required | 18:23 |
fmailhot | Cool, thanks. I'll have a look at that. And if you want to implement multiclass LR, I'd certainly be happy about it! :D (obviously I understand that you & other contributors are busy, though) | 18:24 |
blackburn_ | fmailhot: well I can find time if I know what to do - usually I am a little lost | 18:24 |
blackburn_ | fmailhot: IIRC basic regression is just two lines patch for that tree one | 18:25 |
fmailhot | OK, cool...I'll do some experimenting with that and come back w/ questions as they arise. Thanks again! | 18:25 |
blackburn_ | fmailhot: I just checked - yes I'd have to change regularizer | 18:28 |
blackburn_ | fmailhot: but as it would be highly experimental I give no warranty :D | 18:28 |
fmailhot | :) Fair enough | 18:29 |
blackburn_ | fmailhot: I'll try, test it a little, commit and let you know once I get home | 18:30 |
blackburn_ | so 2-3 hours | 18:30 |
fmailhot | No need to knock yourself out...it's only 10:30AM here in California. | 18:31 |
fmailhot | And tomorrow is fine, too. | 18:31 |
fmailhot | Thanks very much for this! | 18:31 |
blackburn_ | fmailhot: I will forget in other case :D | 18:31 |
blackburn_ | fmailhot: 21:31 here heh | 18:31 |
fmailhot | Ah...heh. OK then. | 18:32 |
blackburn_ | fmailhot: I am from that snowy dark country | 18:33 |
blackburn_ | fmailhot: https://github.com/shogun-toolbox/shogun/blob/master/src/shogun/lib/slep/slep_mc_tree_lr.cpp solver is pretty compact | 18:34 |
fmailhot | Cool, thanks. | 18:35 |
-!- blackburn_ [d5578acb@gateway/web/freenode/ip.213.87.138.203] has quit [Quit: Page closed] | 19:10 | |
-!- blackburn1 [~blackburn@83.234.169.204] has joined #shogun | 19:42 | |
blackburn1 | fmailhot: still here? | 20:48 |
-!- fmailhot is now known as fmailhot_lunch | 20:50 | |
fmailhot_lunch | Hi yes, but about to run out for lunch | 20:50 |
blackburn1 | fmailhot_lunch: just wanted to say I started implementing the method | 20:51 |
fmailhot_lunch | Awesome, thanks! | 20:51 |
fmailhot_lunch | Will be back online in ~1hr. But like I said, no crazy rush...tomorrow would be fine. | 20:52 |
-!- fmailhot_lunch [~fmailhot@c-24-10-116-3.hsd1.ca.comcast.net] has quit [Quit: Leaving.] | 20:53 | |
blackburn1 | I know :) I just don't want to get back to it later | 20:55 |
blackburn1 | fmailhot: see following commits - python example inside | 21:32 |
-shogungit:#shogun- [shogun] lisitsyn pushed 2 new commits to master: https://github.com/shogun-toolbox/shogun/compare/7c1fcabf7d55...3222c06cb214 | 21:32 | |
-shogungit:#shogun- shogun/master 1526e97 Sergey Lisitsyn: Introduced multiclass logistic regression | 21:32 | |
-shogungit:#shogun- shogun/master 3222c06 Sergey Lisitsyn: Added python example of multiclass logistic regression | 21:32 | |
-shogungit:#shogun- [shogun] lisitsyn pushed 1 new commit to master: https://github.com/shogun-toolbox/shogun/commit/5fa8df48872f9a7bc551066f21afac5b49a02c03 | 21:35 | |
-shogungit:#shogun- shogun/master 5fa8df4 Sergey Lisitsyn: Updated reference to submodule tutorial | 21:35 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:57 | |
shogun-buildbot | build #615 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/615 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 22:17 |
-!- fmailhot [~fmailhot@ip-64-134-230-233.public.wayport.net] has joined #shogun | 22:44 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 22:46 | |
blackburn1 | fmailhot: feel free to try | 22:47 |
fmailhot | Cool, thanks...caught up in other things at the moment, but will definitely have a look. | 22:48 |
blackburn1 | fmailhot: please let us know if it is working for you - sorry don't have much time for thorough testing | 22:49 |
fmailhot | No worries, I'll definitely come back w/ a report tomorrow, Friday at the latest. Thanks again very much! | 22:50 |
blackburn1 | fmailhot: are you a researcher btw? | 22:50 |
fmailhot | industry | 22:50 |
blackburn1 | oh that's even more cool | 22:50 |
fmailhot | But doing "research" for the moment... | 22:51 |
fmailhot | Figuring out which libs are best for us | 22:51 |
blackburn1 | hah I see - we need feedback so if you 'choose' us and have time share your ideas | 22:51 |
fmailhot | Sure thing...I'll give you guys feedback either way. | 22:52 |
blackburn1 | thanks | 22:52 |
shogun-buildbot | build #616 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/616 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 22:52 |
blackburn1 | whoops | 22:52 |
blackburn1 | okayy | 22:52 |
ptizoom | snow and sand | 22:53 |
blackburn1 | ptizoom: oh it is you with your poetry :) | 22:53 |
ptizoom | when industry of silicates spread dust over the infinite snowy steps of the east | 22:55 |
blackburn1 | I have no idea what you're talking about :D | 22:56 |
ptizoom | me too | 22:56 |
-shogungit:#shogun- [shogun] lisitsyn pushed 1 new commit to master: https://github.com/shogun-toolbox/shogun/commit/276a2ff8a85cadb935f5740603d45b19ef2d116d | 22:57 | |
-shogungit:#shogun- shogun/master 276a2ff Sergey Lisitsyn: Made multiclass logistic regression python example safe when shogun is compiled without eigen3 | 22:57 | |
* fmailhot thinks he's stepped into a surrealist IRC performance art piece... | 22:58 | |
ptizoom | I see shogun is going well...I try to learn it also bit by bit... there are lots of things in it, incredible. | 22:59 |
blackburn1 | just a few algorithms | 23:00 |
ptizoom | u r 2 modest | 23:00 |
ptizoom | actually I wish there was already a group like this in the nooties... it would have simplified research no end... | 23:02 |
ptizoom | whish | 23:02 |
blackburn1 | not really sure what you mean | 23:03 |
blackburn1 | I failed turing test you know | 23:04 |
ptizoom | well, those ML are wonderful to organise trial and errors...and find converging algo...n'est ce pas? | 23:08 |
ptizoom | and with a package like this, it gives no excuses, not to try... | 23:08 |
blackburn1 | well I see ML quite stupid actually :D | 23:08 |
ptizoom | that the beauty! | 23:09 |
ptizoom | stupid and beautiful | 23:09 |
ptizoom | how many ll in hallowin? | 23:10 |
blackburn1 | come on, what's that again :D | 23:11 |
ptizoom | made pictures and need to label a directory with this name...ok, no help from this channel... lets gg it. | 23:12 |
ptizoom | 2ll and 2ee.... | 23:14 |
-!- blackburn1 [~blackburn@83.234.169.204] has quit [Quit: Leaving.] | 23:24 | |
shogun-buildbot | build #617 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/617 | 23:39 |
-!- fmailhot [~fmailhot@ip-64-134-230-233.public.wayport.net] has left #shogun [] | 23:45 | |
--- Log closed Thu Nov 01 00:00:17 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!