--- Log opened Thu May 19 00:00:11 2011 | ||
@sonney2k | serialhex, hmmhh but this is w/ make distclean before? | 00:20 |
---|---|---|
serialhex | what? the java error? no, it happened when i just said 'make' i can try that in a bit, but i'm still compiling (this time with just ruby & python modular) | 00:22 |
@sonney2k | that shouldn't happen | 00:23 |
@sonney2k | I would suggest to run git clean -dfx | 00:23 |
@sonney2k | to start off fresh | 00:23 |
@sonney2k | (will remove all files not in git) | 00:23 |
CIA-31 | shogun: Soeren Sonnenburg master * rddf2879 / src/libshogun/base/SGObject.h : | 00:24 |
CIA-31 | shogun: Revert "potentially fix a gcc 4.6 compilation error" | 00:24 |
CIA-31 | shogun: This reverts commit 0fe924eca47c3ff5f7c88cac489c2850011d4aec. - http://bit.ly/ihtfGL | 00:24 |
CIA-31 | shogun: Soeren Sonnenburg master * r4c16c19 / (2 files in 2 dirs): introduce SG_UNREF_NOT_NULL to compilation error - http://bit.ly/k2jKSd | 00:24 |
@sonney2k | serialhex, ^^ btw | 00:24 |
@sonney2k | I hope this fixes it | 00:24 |
serialhex | ooh, ok, will cancel & re-try! | 00:25 |
@sonney2k | anyway I will have to sleep now | 00:25 |
@sonney2k | l8r | 00:25 |
serialhex | later, i'll e-mail you with the results sonney2k | 00:26 |
-!- blackburn [~qdrgsm@188.168.4.250] has quit [Quit: Leaving.] | 00:34 | |
-!- ThePing [~phycho@174.127.64.107] has joined #shogun | 02:27 | |
-!- ThePing [~phycho@174.127.64.107] has left #shogun [] | 02:27 | |
-!- sploving [~root@124.16.139.196] has quit [Remote host closed the connection] | 04:51 | |
-!- sonney2k [~shogun@87.118.92.43] has quit [Ping timeout: 258 seconds] | 05:31 | |
-!- sonney2k [~shogun@7nn.de] has joined #shogun | 05:32 | |
-!- mode/#shogun [+o sonney2k] by ChanServ | 05:32 | |
-!- blackburn [~qdrgsm@188.168.5.64] has joined #shogun | 09:21 | |
-!- blackburn [~qdrgsm@188.168.5.64] has quit [Quit: Leaving.] | 10:41 | |
-!- heiko [~heiko@infole-06.uni-duisburg.de] has joined #shogun | 12:08 | |
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has quit [Ping timeout: 260 seconds] | 13:17 | |
-!- sploving [~root@124.16.139.196] has joined #shogun | 13:54 | |
sploving | sonney2k, aroud? | 13:54 |
sploving | I could map the jblas, DoubleMatrix | 13:55 |
sploving | I will push it right now | 13:55 |
-!- heiko [~heiko@infole-06.uni-duisburg.de] has quit [Ping timeout: 258 seconds] | 13:58 | |
@sonney2k | sploving, great! | 14:23 |
@sonney2k | sploving, I suggest to continue with a list of strings | 14:23 |
sploving | sonney2k, I pushed it | 14:23 |
sploving | sonney2k, does SGstring work now? | 14:24 |
@sonney2k | sploving, it will be SGStringList | 14:24 |
@sonney2k | look at DataType.h I introduced this new type | 14:24 |
sploving | I think you could give me python example. like typemap and how to use it | 14:25 |
@sonney2k | will implement some example for it tonight - so would be great if you could start drafting the typemap | 14:25 |
sploving | then I could speed up | 14:25 |
@sonney2k | sploving, just look at the data type and what is currently there from the python side | 14:25 |
sploving | sonney2k, when considering support string, we should not consider jblas, as it does not have it | 14:29 |
@sonney2k | sploving, correct | 14:29 |
sploving | we just consider the java side. | 14:29 |
sploving | OK. then it will be easy | 14:29 |
@sonney2k | sploving, I didn't do python yet but it is easy given that we now know how things work | 14:43 |
sploving | sonney2k, could you give me a test example | 14:44 |
sploving | so I could have a try | 14:44 |
@sonney2k | sploving, not yet | 14:44 |
@sonney2k | hopefully tonight | 14:44 |
@sonney2k | so draft the code only for now and tomorrow you can test | 14:45 |
sploving | sonney2k, it does not matter. I have staring the computer for a whole day. I need a rest | 14:46 |
sploving | sonney2k, SGString<T>, does it mean the string could be double | 14:48 |
sploving | int, and other type? | 14:48 |
sploving | not only char? | 14:48 |
sploving | char * | 14:48 |
-!- sploving [~root@124.16.139.196] has left #shogun [] | 15:07 | |
serialhex | sonney2k: the fix seems to work, now i have to figure out why it's not finding my ruby.h file :-/ | 17:32 |
@mlsec | sonney2k: Alive? | 17:37 |
serialhex | i think he's gone the way of the dodo mlsec :P | 17:40 |
-!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has quit [Remote host closed the connection] | 17:41 | |
serialhex | wtf?? bettyboo left?!!?!? | 17:41 |
@mlsec | nono | 17:41 |
@mlsec | just some internal check | 17:41 |
serialhex | ahh, i see | 17:41 |
@mlsec | betty, now has captured roughly 10k sentences. | 17:42 |
@mlsec | so i need to filter stuff a little. | 17:42 |
serialhex | nice | 17:42 |
serialhex | btw, she's pretty spiffy, i dont know if yo've read some of the logs but she's eerily accurate in her responses | 17:42 |
@mlsec | hehe | 17:43 |
-!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has joined #shogun | 17:44 | |
-!- mode/#shogun [+o bettyboo] by ChanServ | 17:44 | |
@mlsec | it's no black magic. just randomization and simple pattern matching | 17:44 |
serialhex | cool cool | 17:44 |
@mlsec | however, i had time ... well, there is plenty of room for extensions | 17:44 |
@mlsec | if i had | 17:45 |
serialhex | are you going to open source her? | 17:45 |
@mlsec | no. | 17:45 |
serialhex | hey bettyboo, how are you?? | 17:45 |
@bettyboo | serialhex: and assume that cov and mean (together with their sizes) are set as member varialbes | 17:45 |
serialhex | oh, ok | 17:45 |
@mlsec | it's just 240 lines of tcl code | 17:45 |
serialhex | damn | 17:45 |
* serialhex thought it would be more complex than that | 17:46 | |
@mlsec | take 240 lines of code + 9000 captured messages | 17:46 |
@mlsec | and you have a bot | 17:46 |
@mlsec | one great addition that is currently missing in betty, is the ability to track conversations | 17:47 |
@bettyboo | mlsec: how peopleless it is.. | 17:47 |
serialhex | yeah, that would make her even more spooky :D | 17:47 |
@mlsec | in irc it is often not difficult to detect and observe conversations due to nick names in the messages | 17:47 |
@mlsec | using conversations could help to get a better answer/response behavior | 17:48 |
serialhex | that makes sense | 17:48 |
@mlsec | so an obvious test for the lack of this ability is this: | 17:48 |
@mlsec | betty: how are you? | 17:48 |
@bettyboo | mlsec: hey how are you?? | 17:48 |
serialhex | lol | 17:48 |
@mlsec | hehe. it's more matching | 17:49 |
serialhex | yeah, you could do something that matches common responses to Q's like that | 17:49 |
@mlsec | yeah. in principle, you can also automatically detect questions; even if the question mark is missing or the structure is not correct, e.g. you are an atom? | 17:50 |
@mlsec | i have been doing this with a colleague some years ago. tools for parsing and processing natural language text are pretty advanced nowadays | 17:51 |
serialhex | cool cool | 17:51 |
serialhex | yeah i saw a internet chatbot (i forget the url) that was kind of interesting | 17:51 |
serialhex | the one thing that seems to be missing in all the ones i've seen/heard of is the ability to start a conversation on it's own, i'm sure it's easy to do random strings, but something more advanced is probably much harder | 17:53 |
@mlsec | yeah | 17:53 |
@mlsec | brings us back to conversations. if we would track them, we could easily pick a first message to start a chat | 17:54 |
serialhex | cool | 17:54 |
-!- alesis-novik [~alesis@188.74.87.84] has quit [Quit: I'll be Bach] | 18:32 | |
-!- blackburn [~qdrgsm@188.168.4.67] has joined #shogun | 19:34 | |
serialhex | oi blackburn!! | 19:48 |
blackburn | hi! | 19:48 |
blackburn | how are ya? | 19:49 |
serialhex | pretty good, yourself? | 19:49 |
blackburn | fine, just waked up :D | 19:49 |
serialhex | wtf?? it's got to be like 11pm or something there? [my cool clocks arent set up :( ] | 19:50 |
blackburn | aha, fell asleep about 7pm and waked up 9pm :D | 19:51 |
serialhex | ahh, cool catnap :D | 19:53 |
@bettyboo | rahaha | 19:53 |
blackburn | http://nvie.com/img/2009/12/Screen-shot-2009-12-24-at-11.32.03.png | 20:26 |
@sonney2k | serialhex, did my patch fix the compile issue? | 20:44 |
@sonney2k | ahh ok good | 20:46 |
blackburn | sonney2k: added examples for both prc and roc | 20:48 |
serialhex | sonney2k: i guess you saw, i'm going to try and fix the problem with ruby not being compiled but shogun itself seems to compile fine | 20:51 |
serialhex | bbl, doin some house work | 20:52 |
@sonney2k | blackburn, nice :) | 20:58 |
blackburn | sonney2k: thinked a lot about which classifiers to choose :D LDA+SVM seems to be ok? | 20:59 |
@sonney2k | blackburn, you could illustrate that even w/o classifiers | 20:59 |
@sonney2k | e.g. just use some random data | 21:00 |
blackburn | I know, but this way could be more 'real' | 21:00 |
@sonney2k | sure | 21:00 |
blackburn | it shows that svm is better than lda :D | 21:00 |
@sonney2k | blackburn, well depends | 21:20 |
@sonney2k | single hand typing sucks... | 21:21 |
blackburn | why single? | 21:21 |
@sonney2k | one arm baby | 21:22 |
blackburn | such a kung-fu :D | 21:22 |
@bettyboo | blackburn, hoho! | 21:22 |
@sonney2k | blackburn, now 2 arms :) | 21:24 |
@sonney2k | I looked at your examples | 21:25 |
@sonney2k | you should create one without figures | 21:25 |
@sonney2k | too | 21:25 |
blackburn | okay | 21:25 |
@sonney2k | otherwise we have no testsuite for this | 21:25 |
@sonney2k | and not just roc/prc | 21:25 |
@sonney2k | but all the measures | 21:25 |
@sonney2k | you could do the for random data for this purpose | 21:26 |
blackburn | i'll do it a bit later | 21:26 |
@sonney2k | slighly correlated with label ... | 21:26 |
blackburn | ah, i'd rather do it now hehe | 21:28 |
CIA-31 | shogun: Sergey Lisitsyn master * r3bec041 / (2 files): Added graphical examples for PRC and ROC - http://bit.ly/mq1v1y | 21:28 |
CIA-31 | shogun: Soeren Sonnenburg master * r064cdc5 / (2 files): Merge branch 'master' of git://github.com/SergeyLisitsyn/shogun - http://bit.ly/imDeBC | 21:28 |
@sonney2k | blackburn, go ahead | 21:28 |
-!- ameerkat [~ameerkat@184-98-140-155.phnx.qwest.net] has joined #shogun | 21:28 | |
@sonney2k | blackburn, figures look cool. one way to improve would be to plot both rocs in one figure and make one red-ish transparent and one blueish | 21:29 |
blackburn | sonney2k: they are almost equal sometimes, will it be informative? | 21:30 |
@sonney2k | blackburn, I usually plot roc curves just as lines | 21:31 |
@sonney2k | and for many classifiers in one figure | 21:31 |
@sonney2k | this way I can see when which classifier is better | 21:31 |
blackburn | sonney2k: I see | 21:33 |
@sonney2k | anyway nothing mandatory - figures look nice | 21:34 |
blackburn | okay | 21:35 |
blackburn | oops | 21:40 |
CIA-31 | shogun: Soeren Sonnenburg master * rc852a03 / (2 files in 2 dirs): | 21:44 |
CIA-31 | shogun: Merge branch 'master' of git://github.com/sploving/shogun | 21:44 |
CIA-31 | shogun: Conflicts: | 21:44 |
CIA-31 | shogun: src/java_modular/swig_typemaps.i - http://bit.ly/jTwPPr | 21:44 |
blackburn | sonney2k: Labels(numpy.random.randn(1,N)) caused a memory corruption :D | 21:45 |
blackburn | because it is [[....]] but not [...] | 21:46 |
blackburn | sonney2k: | 22:06 |
blackburn | evaluator = ContingencyTableEvaluation() | 22:06 |
blackburn | assert evaluator.evaluate(predicted_labels,ground_truth_labels) == base_evaluator.get_accuracy() | 22:06 |
blackburn | good for test? | 22:06 |
blackburn | tfuu! | 22:06 |
blackburn | evaluator = AccuracyMeasure() | 22:07 |
blackburn | assert evaluator.evaluate(predicted_labels,ground_truth_labels) == base_evaluator.get_accuracy() | 22:07 |
blackburn | or just compute it and nothing? | 22:07 |
@sonney2k | just compute it and return it | 22:08 |
@sonney2k | like the example you have written about a month ago | 22:08 |
blackburn | sonney2k: there are helper classes: AccuracyMeasure, etc | 22:09 |
@sonney2k | blackburn, why don't you use rand(10) ? | 22:09 |
blackburn | sonney2k: where? | 22:09 |
@sonney2k | Labels(numpy.random.randn(1,N)) | 22:09 |
@sonney2k | why the 1, there? | 22:09 |
blackburn | i fixed it already | 22:09 |
@sonney2k | ok | 22:10 |
blackburn | sonney2k: so about helpers | 22:10 |
blackburn | should I test it in one .py? | 22:10 |
blackburn | because all of them are similar, it could be better to test it at once | 22:10 |
@sonney2k | blackburn, I would do it like you did in C++ | 22:14 |
blackburn | eh? | 22:14 |
@sonney2k | all the contingency table based ones in one file | 22:14 |
blackburn | ah | 22:14 |
@sonney2k | the others roc / prc in different | 22:14 |
blackburn | yeap I did it this way for now | 22:14 |
@sonney2k | fine | 22:16 |
blackburn | sonney2k: done | 22:28 |
@sonney2k | blackburn, I am currently writing some helper functions for strings / sparse matrix | 22:28 |
@sonney2k | to use our new datatypes | 22:29 |
blackburn | nice | 22:29 |
@sonney2k | blackburn, please seed the random number generator such that this is reproducible | 23:04 |
blackburn | sonney2k: do you think it is needed for numpy.random? | 23:07 |
@sonney2k | sonney2k, yes it is needed | 23:07 |
blackburn | okay | 23:07 |
@sonney2k | numpy.random.seed(17) :) | 23:08 |
blackburn | sonney2k: done | 23:09 |
CIA-31 | shogun: Soeren Sonnenburg master * r81c627f / (5 files): Merge branch 'master' of git://github.com/SergeyLisitsyn/shogun (+6 more commits...) - http://bit.ly/jbqwSA | 23:11 |
@sonney2k | merged | 23:11 |
@sonney2k | thanks! | 23:11 |
blackburn | sonney2k: have you seen this scheme? http://nvie.com/img/2009/12/Screen-shot-2009-12-24-at-11.32.03.png | 23:12 |
blackburn | we could use something similar | 23:13 |
@sonney2k | blackburn, you mean a development branch? | 23:14 |
serialhex | i suggested that weeks ago blackburn and no one listened to me!!! :P but if you have more luck convincing ppl then thats awesome! (i thinkit's pretty spiffy myself!) | 23:14 |
blackburn | hotfixes + develop + features | 23:14 |
blackburn | I'm already using branches for lle and mds | 23:14 |
serialhex | https://github.com/nvie/gitflow is the main link & http://jeffkreeftmeijer.com/2010/why-arent-you-using-git-flow/ is a spiffy video | 23:15 |
@sonney2k | blackburn, sure but the problem is how to do that across people - I mean e.g. the SGVector business is something that sploving, mikio and I need | 23:17 |
@sonney2k | and heiko contributed to | 23:17 |
blackburn | sonney2k: just the naming issue - i.e I should use mds,lle for my project work, hotfixes for fixes :) and some other branch to small additions | 23:19 |
blackburn | e.g. sploving could do it in java branch or so | 23:19 |
@sonney2k | I guess we could have created a branch for that and only merged this back to master once everything is done | 23:19 |
@bettyboo | ho sonney2k | 23:19 |
@sonney2k | blackburn, you can name your branches as you like... workflowwise the only thing I prefer is to merge small changes often | 23:21 |
@sonney2k | what is different in this release model is that we do (joint) development in master only | 23:25 |
blackburn | yeap | 23:26 |
@sonney2k | bigger features in some feature branches yes (smaller ones directly in master) | 23:26 |
@sonney2k | we only tag releases so far | 23:26 |
@sonney2k | no branch for releases + hotfixes | 23:26 |
@sonney2k | (yet) | 23:26 |
blackburn | may be some more branches could be better to comprehend, well idk if it would be better | 23:26 |
@sonney2k | btw, I used to do this back in the days when we uses cvs for shogun | 23:27 |
@sonney2k | but then we never needed the release branch | 23:27 |
@sonney2k | I guess now it makes more sense considering shogun's size | 23:27 |
@sonney2k | anyway for isolated features all these branches are very helpful | 23:28 |
@sonney2k | for such intrusive change as the SGVector* one it does not help - this change stays a nightmare | 23:28 |
blackburn | yeap | 23:29 |
@sonney2k | blackburn, though we made good progress... python typemaps are there more or less and heiko did the serialization support already | 23:30 |
blackburn | sonney2k: 3 days to start of gsoc :) | 23:30 |
@sonney2k | it all needs testing now, support for r_modular/octave_modular and more documentation / changes in the classes | 23:30 |
@sonney2k | blackburn, yeah that will also mean weekly student reports... | 23:31 |
blackburn | I would report on my progress on this sunday | 23:31 |
@sonney2k | but heh I hope some projects really get rolling | 23:31 |
@sonney2k | well 1 week after gsoc starts is sufficient :) | 23:32 |
blackburn | I hope to finish ~1/2 of my project before my exams | 23:32 |
blackburn | and another 1/2 right before mid-term :) | 23:32 |
@sonney2k | I wish you best of luck! | 23:32 |
blackburn | after mid-term I want to make a paper about dimreduction in shogun and some cleanup/support and so on | 23:33 |
@sonney2k | blackburn, I really hope that around mid-term we have the SGVector* transition finished and also the renaming of CClassifier -> CMethod | 23:34 |
blackburn | CMachine? | 23:34 |
@sonney2k | or CMachine | 23:35 |
blackburn | I think CMethod could be a nightmare | 23:35 |
@sonney2k | probably better - we do MACHINE learning | 23:35 |
blackburn | e.g. | 23:35 |
blackburn | // method of method that methods the method | 23:35 |
blackburn | :D | 23:35 |
@sonney2k | I wish I had more time to do these transitions now - all I have is 1hr/day max currently | 23:36 |
@sonney2k | speaking of which - I need to sleep | 23:36 |
blackburn | it just like name some class CClass | 23:36 |
@sonney2k | cu you around and serialhex good luck with ruby | 23:37 |
@bettyboo | bye | 23:37 |
blackburn | see you | 23:37 |
serialhex | no | 23:37 |
@sonney2k | blackburn, CGoodNight! | 23:37 |
serialhex | np i mean :P | 23:37 |
serialhex | nite! | 23:37 |
blackburn | sonney2k: new CHaveANiceDream() | 23:37 |
serialhex | lol | 23:37 |
blackburn | or more java-style | 23:37 |
@sonney2k | CClosingEyes() | 23:37 |
blackburn | InitialContext ic = new InitialContext(); | 23:37 |
serialhex | http://emergency.cdc.gov/socialmedia/zombies_blog.asp we're prepared 4 zombie apocalypse, r u?? | 23:38 |
@sonney2k | b= new CBrain(); delete b; | 23:38 |
serialhex | lol sonney2k | 23:38 |
blackburn | Object ref = ic.lookup("HaveANiceDreamFactory") | 23:38 |
blackburn | HaveANiceDreamFactoryRemoteHome factory = (HaveANiceDreamFactoryRemoteHome) PortableRemoteObject.narrow(ref, HaveANiceDreamFactoryRemoteHome.class); | 23:38 |
blackburn | HaveANiceDreamFactoryRemote = factory.create(); | 23:39 |
blackburn | :D | 23:39 |
blackburn | HaveANiceDream dream = factoryRemote.create(); | 23:40 |
blackburn | fuck I have to sleep now | 23:40 |
serialhex | soeren.sleep :dream => :good_ones | 23:40 |
blackburn | serialhex: looks commonlispy | 23:40 |
serialhex | it's ruby | 23:40 |
serialhex | with a hash arg :D | 23:40 |
blackburn | ah, in ruby they use colons too | 23:40 |
serialhex | the colons are for symbols, which are lightweight & immutable strings | 23:41 |
blackburn | it is the lisp way | 23:41 |
serialhex | they're spiffy for use in hashes & other things | 23:41 |
serialhex | cool | 23:41 |
blackburn | and of course it is clojure way too :) | 23:42 |
serialhex | yeah, cause coujure >= lisp :P | 23:42 |
blackburn | I played with clojure a little but didn't learnt anything cool for some reason | 23:43 |
blackburn | anyway now I use only three languages: python, java and C++ :) | 23:44 |
serialhex | hmm, i need to get thru this book that will teach me the basics for 7 langs | 23:44 |
blackburn | I think it is useless :) | 23:45 |
serialhex | yeah me == ruby :P | 23:45 |
serialhex | with << C/C++ | 23:45 |
blackburn | anyone will forget these things about not-needed-languages | 23:45 |
blackburn | I have used a common lisp for a while but now cannot remember anything - I just don't need it | 23:46 |
blackburn | same thing with physics :D | 23:46 |
serialhex | yeah, if u dont use it then u wont remember it | 23:46 |
blackburn | well it is a bit late | 23:47 |
blackburn | going sleep just like Soeren did | 23:47 |
serialhex | ok | 23:47 |
blackburn | serialhex: have a nice eh.. evening/ | 23:47 |
blackburn | ? | 23:47 |
blackburn | :) | 23:47 |
serialhex | yeah, thanks, nite blackburn! | 23:47 |
blackburn | see you | 23:47 |
-!- blackburn [~qdrgsm@188.168.4.67] has quit [Quit: Leaving.] | 23:48 | |
--- Log closed Fri May 20 00:00:13 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!