--- Log opened Sat Mar 16 00:00:13 2013 | ||
-!- FSCV [~FSCV@206.225.135.24] has quit [Ping timeout: 276 seconds] | 00:02 | |
-!- FSCV [~FSCV@108.60.131.13] has joined #shogun | 00:04 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 264 seconds] | 00:57 | |
-!- FSCV [~FSCV@108.60.131.13] has quit [Quit: Leaving] | 00:59 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 04:24 | |
-!- hoijui [~hoijui@dslb-088-074-124-160.pools.arcor-ip.net] has joined #shogun | 07:34 | |
@sonney2k | blackburn, we still have crashes. as soon as make prepare-release works we can go further | 08:56 |
---|---|---|
@sonney2k | blackburn, feel free to run that your own and fix the crashers | 08:56 |
@sonney2k | blackburn, I don't have time today to work on it so any progress helps | 09:05 |
blackburn | sonney2k: okay | 09:09 |
wiking | check this plz and if u agree then let's merge it: https://github.com/shogun-toolbox/shogun/pull/941 | 10:32 |
-!- hoijui [~hoijui@dslb-088-074-124-160.pools.arcor-ip.net] has quit [Ping timeout: 256 seconds] | 10:34 | |
-!- votjak [~votjak@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun | 10:56 | |
wiking | check this plz and if u agree then let's merge it: https://github.com/shogun-toolbox/shogun/pull/941 | 11:07 |
wiking | oh sorry | 11:07 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 11:13 | |
n4nd0 | hey blackburn | 11:28 |
n4nd0 | where can I find the pictures that compose https://dl.dropbox.com/u/10139213/shogun/pics.png? | 11:28 |
blackburn | n4nd0: hey | 11:46 |
blackburn | oh let me find it | 11:46 |
blackburn | wiking: so -a is more widespread than --append? | 11:47 |
wiking | yes | 11:53 |
wiking | only linux has --apend | 11:54 |
wiking | *--append | 11:54 |
blackburn | wiking: let it merge then | 11:55 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 11:55 | |
shogun-notifier- | shogun: Viktor Gal :master * c758bf6 / Makefile: https://github.com/shogun-toolbox/shogun/commit/c758bf6f045a7e169c5696fef4e6c616c0c7d6e3 | 11:55 |
shogun-notifier- | shogun: use -a for tee as that's more widespread among OSes | 11:55 |
shogun-notifier- | shogun: Sergey Lisitsyn :master * 3d7cf1a / Makefile: https://github.com/shogun-toolbox/shogun/commit/3d7cf1a359a310f1e22eff5334b9fb6a3b684ea8 | 11:55 |
shogun-notifier- | shogun: Merge pull request #941 from vigsterkr/master | 11:55 |
shogun-notifier- | shogun: | 11:55 |
shogun-notifier- | shogun: use -a for tee as that's more widespread among OSes | 11:55 |
blackburn | n4nd0: I found a script that produced that but lost images :D | 11:56 |
n4nd0 | blackburn: hehe ok, I will play around with faces then | 11:59 |
shogun-buildbot_ | build #807 of deb2 - static_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/807 | 12:04 |
blackburn | n4nd0: I will prepare road signs HOG features | 12:09 |
blackburn | n4nd0: but we'd need your machine to handle that :D | 12:10 |
n4nd0 | blackburn: sure, no problem | 12:25 |
n4nd0 | blackburn: is it to put more tests in tapkee? | 12:25 |
blackburn | n4nd0: no, it is rather example | 12:29 |
blackburn | n4nd0: yikes, the data is 1.4 gb | 12:30 |
n4nd0 | blackburn: wow how many feature vectors and dimension? | 13:27 |
blackburn | n4nd0: 39209 of 3200d | 14:08 |
n4nd0 | wow | 14:15 |
blackburn | n4nd0: I'd reduce it to somewhat 400d I think | 14:20 |
blackburn | or it gets pretty hard to embed in reasonable time | 14:21 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 14:55 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 16:11 | |
wiking | sonney2k: around? | 17:32 |
-!- wiking [~wiking@info2k1.hu] has quit [Quit: leaving] | 17:45 | |
-!- hoijui [~hoijui@dslb-088-074-124-160.pools.arcor-ip.net] has joined #shogun | 18:36 | |
-!- heiko [~heiko@5ac1f59f.bb.sky.com] has joined #shogun | 18:42 | |
heiko | blackburn: hi | 18:43 |
blackburn | heiko: hey | 18:43 |
heiko | just reading MulticlassSVM header file | 18:43 |
heiko | we really should not remove warnings if documentation is missing by putting /** @brief MulticlassSVM */ | 18:43 |
heiko | same for the methods :) | 18:43 |
blackburn | heiko: heh right | 18:44 |
heiko | thats way too lazy :D | 18:44 |
blackburn | heiko: I forgot about that bug report | 18:44 |
heiko | blackburn: no bug here | 18:44 |
blackburn | running prepare-release now | 18:44 |
blackburn | heiko: what is wrong? | 18:44 |
heiko | the guy just wrote bad code | 18:44 |
blackburn | ahh | 18:44 |
heiko | uninitialised memory etc | 18:44 |
heiko | everything works | 18:44 |
heiko | Ill send him a nice example back | 18:44 |
heiko | and add to libshogun | 18:44 |
blackburn | well we are who to blame for that people can write bad code :) | 18:45 |
heiko | blackburn: you are totally right | 18:46 |
heiko | but he did not even init his label memory | 18:46 |
heiko | so there is no way we can avoid that in c++ | 18:46 |
blackburn | heiko: I am not sure about that - may be there is a way | 18:46 |
-!- votjak [~votjak@host-46-241-3-209.bbcustomer.zsttk.net] has quit [] | 19:00 | |
blackburn | heiko: dst.template triangularView<Upper>() = src.template triangularView<Upper>(); | 19:03 |
blackburn | heiko: ^ that's valid C++ | 19:03 |
blackburn | :D | 19:03 |
blackburn | sometimes I wanna cry | 19:03 |
heiko | blackburn: ? :) | 19:03 |
heiko | what happens if you do it | 19:03 |
blackburn | heiko: dst.template[Space here]<Upper>() | 19:04 |
blackburn | heiko: crazy C++ | 19:04 |
heiko | blackburn: am I allowed to commit stuff? | 19:06 |
blackburn | heiko: yes, for a year already :D | 19:06 |
heiko | blackburn: you are a scherzkeks | 19:06 |
heiko | I mean does it interfer with your release work? | 19:07 |
blackburn | heiko: ah no, I just test if it works | 19:07 |
heiko | ok | 19:07 |
heiko | blackburn: just ran into another multiclass svm bug :( | 19:08 |
blackburn | heiko: scherzkeks is kind of 'joker'? | 19:08 |
heiko | blackburn: yes :) | 19:08 |
blackburn | heiko: what is da bug? | 19:09 |
heiko | blackburn: I will add the example | 19:11 |
heiko | then it will fail | 19:11 |
heiko | and you are forced to solve it :D | 19:11 |
heiko | REQUIRE(set_svm(idx, svm),"SVM set failed") | 19:11 |
heiko | multiclasslibsvm | 19:11 |
heiko | line 194 fails | 19:11 |
heiko | depending on inpu data | 19:11 |
heiko | another really nice bug | 19:11 |
blackburn | heiko: ahh there was some issue with that | 19:12 |
heiko | blackburn: | 19:15 |
heiko | it only occurs with multiple classes | 19:16 |
heiko | Ill add example | 19:16 |
heiko | and you have to increase number of classes from two to 3 or something to see what fails | 19:16 |
blackburn | heiko: btw would be nice to finally release asap or it gets forever | 19:16 |
heiko | blackburn: agreed | 19:16 |
heiko | this bug would be nice to fix | 19:16 |
heiko | (also mention it in NEWS) | 19:17 |
blackburn | heiko: ok | 19:17 |
heiko | and the guy from the mailing list would be happy | 19:17 |
heiko | blackburn https://github.com/shogun-toolbox/shogun/pull/942 | 19:19 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 19:19 | |
shogun-notifier- | shogun: Heiko Strathmann :master * 51e5322 / examples/undocumented/libshogun/ (2 files): https://github.com/shogun-toolbox/shogun/commit/51e53229e94afb57b898ac86931ab5f6238fc9bb | 19:19 |
shogun-notifier- | shogun: added a modelselection example for multiclass lib svm | 19:19 |
shogun-notifier- | shogun: Heiko Strathmann :master * a88be17 / examples/undocumented/libshogun/ (2 files): https://github.com/shogun-toolbox/shogun/commit/a88be17180977f29fa5fee9bbfa58bc87e0eecc3 | 19:19 |
shogun-notifier- | shogun: Merge pull request #942 from karlnapf/master | 19:19 |
shogun-notifier- | shogun: | 19:19 |
shogun-notifier- | shogun: new example for model selection multiclass svm | 19:19 |
heiko | blackburn: set number of dimensions/classes to 3 and you see the problem. If it works, could you permantenly increase it? | 19:20 |
heiko | blackburn: I gotta go now, will write to the mailing list and see you tomoroow | 19:20 |
blackburn | okay | 19:20 |
blackburn | heiko: I'll check | 19:20 |
heiko | blackburn: thanks! | 19:26 |
heiko | blackburn: and a set_autolock method for multiclass would be great | 19:26 |
heiko | or even a lock method :) | 19:26 |
heiko | but thats for later | 19:26 |
heiko | the set_atuolock would prevent the warning that is outputted in the ex. | 19:27 |
heiko | blackburn: ok gotta run now. bye! | 19:27 |
-!- heiko [~heiko@5ac1f59f.bb.sky.com] has left #shogun [] | 19:27 | |
blackburn | bye | 19:27 |
-!- zxtx [~zv@75.59.238.10] has joined #shogun | 20:17 | |
shogun-buildbot_ | build #354 of rpm1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/354 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 20:26 |
@sonney2k | blackburn, any progress towards release? | 20:36 |
blackburn | sonney2k: yeah I'll fix a bug pretty soon | 20:37 |
blackburn | sonney2k: I am not sure what you meant by crashes | 20:37 |
blackburn | I ran prepare-release (takes \infty) and there were no real crashes | 20:37 |
@sonney2k | blackburn, really? | 20:38 |
@sonney2k | it crashed here | 20:38 |
blackburn | sonney2k: what exactly? | 20:38 |
@sonney2k | classifier_perceptron in octave static | 20:38 |
@sonney2k | features_dot_wdfeatures.m | 20:39 |
@sonney2k | classifier_svmocas_objective.m | 20:39 |
blackburn | sonney2k: I tried octave_static examples - all ran | 20:39 |
blackburn | there are a lot of integration test errors, but we can't fix it yet | 20:40 |
@sonney2k | yeah sure not a blocker | 20:40 |
@sonney2k | blackburn, btw make prepare-release takes so long b'coz it compiles shogun for all $LANGs | 20:41 |
@sonney2k | so now you know why I would love to have lower compile times | 20:42 |
blackburn | sonney2k: we'd have to speedup swig for that | 20:43 |
blackburn | :D | 20:43 |
blackburn | [0.993604,1.002475] with alpha=0.050000, mean=0.998039 | 20:43 |
blackburn | sonney2k: minor bug in xval :D | 20:44 |
blackburn | sonney2k: conf interval is not cut by maximum possible value | 20:44 |
blackburn | and we tell users accuracy is in between 0.99 and 1.002 hah | 20:44 |
@sonney2k | blackburn, not swig is the problem... | 20:48 |
blackburn | sonney2k: compilation of classes is pretty fast | 20:48 |
@sonney2k | libshogun is fast | 20:48 |
blackburn | sonney2k: at least getting faster with new versions of compilers | 20:48 |
@sonney2k | but we would need split up version of the swig wrapper | 20:48 |
@sonney2k | blackburn, ohh octave seems to work now - the environment variable needs to be OCTAVE_PATH no longer OCTAVE_LOADPATH | 20:49 |
@sonney2k | *sigh* | 20:49 |
blackburn | heh | 20:49 |
shogun-notifier- | shogun: Sergey Lisitsyn :master * a77b5ba / src/ChangeLog: https://github.com/shogun-toolbox/shogun/commit/a77b5baaa22e07fb8e0b597b4452adce1abefcd7 | 20:51 |
shogun-notifier- | shogun: updated changelog | 20:51 |
shogun-notifier- | shogun: Sergey Lisitsyn :master * fc7f16e / src/ChangeLog: https://github.com/shogun-toolbox/shogun/commit/fc7f16e0ba181fbd2b0c4d27be0cfccf6b55a562 | 20:51 |
shogun-notifier- | shogun: updated changelog | 20:51 |
shogun-notifier- | shogun: Sergey Lisitsyn :master * c77af73 / examples/undocumented/libshogun/ (2 files): https://github.com/shogun-toolbox/shogun/commit/c77af730f686f4d6b29f14b4a647e45c17436160 | 20:51 |
shogun-notifier- | shogun: Merge branch 'master' of github.com:shogun-toolbox/shogun | 20:51 |
shogun-notifier- | shogun: Sergey Lisitsyn :master * ec6e149 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/ec6e14953a4b192a3c9d57ed1a9bb59d39220abc | 20:51 |
shogun-notifier- | shogun: Fixed bug with libsvm 3 classes | 20:51 |
blackburn | argh | 20:51 |
blackburn | sonney2k: yikes, these two commits were produced by prepare-release | 20:52 |
blackburn | I am not sure what changelog is used for though | 20:52 |
blackburn | except gpl stuff | 20:53 |
@sonney2k | blackburn, yeah I know | 20:54 |
@sonney2k | next time git reset --hard | 20:54 |
blackburn | sonney2k: git reset HEAD~1 is enough | 20:55 |
@sonney2k | blackburn, but that is a soft reset then right? and you still have this change lying around | 21:00 |
blackburn | sonney2k: they are erased but next commits I think | 21:00 |
blackburn | sonney2k: argh we failed to release on my bday :D | 21:01 |
blackburn | 56 seconds ago | 21:01 |
@sonney2k | blackburn, when was it? | 21:01 |
blackburn | sonney2k: today for you, yesterday for me | 21:01 |
@sonney2k | blackburn, then congrats. now you are approaching the thirties quickly :D | 21:03 |
blackburn | sonney2k: hah thanks | 21:03 |
blackburn | sonney2k: 8 more years to approach it | 21:03 |
@sonney2k | blackburn, so you can still become professor before that :) | 21:05 |
blackburn | sonney2k: phd in doing useless things :D | 21:06 |
@sonney2k | argh I need a faster machine | 21:06 |
@sonney2k | blackburn, for example machine learning ;) | 21:06 |
blackburn | sonney2k: yeah pretty useless stuff mostly | 21:06 |
@sonney2k | blackburn, so go to guu and cure cancer! | 21:06 |
blackburn | heh | 21:06 |
blackburn | guu sounds funny | 21:07 |
@sonney2k | blackburn, our json serialization is broken :/ | 21:09 |
@sonney2k | serialization_complex_example.lua died | 21:09 |
shogun-notifier- | shogun: Soeren Sonnenburg :master * 0e282c7 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/0e282c746ca3711b46d31496e6351c33663cbcc1 | 21:26 |
shogun-notifier- | shogun: release fixes | 21:26 |
shogun-notifier- | shogun: | 21:26 |
shogun-notifier- | shogun: - disable json serialization for now | 21:26 |
shogun-notifier- | shogun: - rename OCTAVE_LOADPATH -> OCTAVE_PATH | 21:26 |
blackburn | sonney2k: hmm | 21:26 |
blackburn | sonney2k: any reason? | 21:27 |
blackburn | sonney2k: I mean it worked before right? | 21:27 |
@sonney2k | no idea | 21:27 |
@sonney2k | yes | 21:27 |
blackburn | quite strange then | 21:27 |
blackburn | I do not remember any chagnes | 21:27 |
shogun-buildbot_ | build #355 of rpm1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/355 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 21:28 |
@sonney2k | blackburn, I couldn't find any either | 21:28 |
@sonney2k | https://research.microsoft.com/pubs/155552/decisionForests_MSR_TR_2011_114.pdf | 21:35 |
@sonney2k | nice intro to random forests | 21:35 |
-!- hoijui [~hoijui@dslb-088-074-124-160.pools.arcor-ip.net] has quit [Ping timeout: 252 seconds] | 21:36 | |
blackburn | sonney2k: actually I never tried them (shame on me) | 21:37 |
@sonney2k | blackburn, they work as good as SVMs | 21:38 |
blackburn | sonney2k: I beat them on GTSRB | 21:38 |
@sonney2k | I used them to predict whether a molecule is a drug or not back in the days :D | 21:38 |
@sonney2k | GTRSRB? | 21:38 |
blackburn | sonney2k: german traffic sign recognition benchmark | 21:38 |
@sonney2k | ahh | 21:39 |
blackburn | sonney2k: funny thing happened with GTSDB (detection instead of recognition) | 21:39 |
blackburn | sonney2k: they all got 1.0 ROCs | 21:39 |
@sonney2k | blackburn, yeah they cannot provide realvalued outputs | 21:39 |
@sonney2k | it is a majority vote | 21:39 |
@sonney2k | 100 trees... | 21:40 |
@sonney2k | btw, bagged KNN | 21:40 |
@sonney2k | works also very very well | 21:40 |
blackburn | sonney2k: http://benchmark.ini.rub.de/?section=gtsdb&subsection=results | 21:40 |
@sonney2k | no idea if that also has a standard name | 21:40 |
blackburn | sonney2k: never heard about special name for that | 21:40 |
blackburn | sonney2k: bagging is effective thing for anything I guess | 21:41 |
blackburn | efficient | 21:41 |
blackburn | :D | 21:41 |
@sonney2k | blackburn, are you in this table? | 21:41 |
blackburn | sonney2k: I wish I was | 21:42 |
@sonney2k | some ELL_SVM2 won? | 21:42 |
blackburn | sonney2k: I started my research on recognition after GTSRB and I just had no time to do GTSDB | 21:42 |
blackburn | sonney2k: not sure what ELL means | 21:43 |
blackburn | sonney2k: second place is integral channel features with multiple scaling I guess | 21:43 |
blackburn | according to its name | 21:43 |
@sonney2k | blackburn, alright I am off to bed. | 21:44 |
blackburn | sonney2k: good night | 21:44 |
@sonney2k | blackburn, get drunk and celebrate! | 21:44 |
blackburn | sonney2k: I have to finish paper stuff | 21:45 |
@sonney2k | I hope we get prepare release to work | 21:45 |
@sonney2k | blackburn, then get drunk and finish paper stuff :) | 21:45 |
blackburn | sonney2k: 1.5 years of accident writing jmlr paper is crazy | 21:45 |
@sonney2k | blackburn, well you could have had the paper about 1 year ago but you took the long route | 21:46 |
* sonney2k Zzzzz | 21:46 | |
blackburn | sonney2k: good thing is that it is fast now | 21:46 |
blackburn | and being used for real | 21:46 |
blackburn | by chemists though :D | 21:46 |
blackburn | I expected ML/data analysis hah | 21:46 |
-!- zxtx [~zv@75.59.238.10] has quit [Ping timeout: 240 seconds] | 22:06 | |
-!- zxtx [~zv@75-59-238-10.lightspeed.sntcca.sbcglobal.net] has joined #shogun | 22:07 | |
shogun-buildbot_ | build #356 of rpm1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/356 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 22:31 |
-!- zxtx [~zv@75-59-238-10.lightspeed.sntcca.sbcglobal.net] has quit [Ping timeout: 264 seconds] | 23:01 | |
shogun-buildbot_ | build #357 of rpm1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/357 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:35 |
--- Log closed Sun Mar 17 00:00:13 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!