IRC logs of #shogun for Wednesday, 2011-07-13

« Tuesday, 2011-07-12 Index Thursday, 2011-07-14 »
--- Log opened Wed Jul 13 00:00:14 2011
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]03:24
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun03:38
-!- f-x [~user@117.192.220.22] has quit [Ping timeout: 260 seconds]04:24
-!- in3xes_ [~in3xes@180.149.49.230] has joined #shogun07:37
-!- in3xes_ [~in3xes@180.149.49.230] has quit [Quit: Leaving]08:26
CIA-32shogun: Shashwat Lal Das master * r58f3522 / (3 files): Temp: Some debug outputs from svmsgd. - http://bit.ly/qr8xln08:33
CIA-32shogun: Shashwat Lal Das master * r77e0762 / (3 files in 2 dirs): Changed default calibration sample to 1000 vectors in online SGD, and size of ring to be 1024 for StreamingSimpleFeatures. - http://bit.ly/mQGVYY08:33
CIA-32shogun: Shashwat Lal Das master * rc4ec26b / (4 files in 2 dirs):08:33
CIA-32shogun: Revert temp SGD change.08:33
CIA-32shogun: Added OnlineLibLinear using one solver.08:33
CIA-32shogun: Change in StreamingSimpleFeatures, implementing dot() when called with another StreamingDotFeatures object. - http://bit.ly/pjedZ308:33
CIA-32shogun: Soeren Sonnenburg master * rd801a24 / (5 files in 2 dirs):08:33
CIA-32shogun: Merge pull request #188 from frx/streaming_108:33
CIA-32shogun: OnlineLibLinear and some other minor changes - http://bit.ly/oSUcfM08:33
CIA-32shogun: Shashwat Lal Das master * r3c0bb6e / (2 files): Revert temp OnlineSVMSGD change. - http://bit.ly/ncxBeO08:33
@sonney2kalesis-novik, are you around? I haven't heard back from you...08:37
bettyboohello08:37
alesis-novikHello sonney2k, sorry, I wasn't paying attention to IRC. What's up?08:38
@sonney2kalesis-novik, I didn't see your weekly report - or?08:38
@sonney2kso I am naturally wondering if everything is going OK08:38
alesis-novikI sent it not too long ago. A bit late, sorry about that08:39
alesis-novikYeah, I understand. I'm currently working on Split-Merge EM like I discussed with Vojtech.08:39
@sonney2kOK, just please take this reporting more serious. It is the only way for me to see if you are making progress when you don't send any patches (IIRC you haven't this week)08:40
@sonney2kso missing the monthly progress report deadline rings warning bells here.08:40
@sonney2kanyway weitermachen as we would say in germany08:41
alesis-novikActually sonney2k, should I add the usage examples I made in python somewhere (even though the script need a bit cleaning, it works well)?08:43
@sonney2ksure - clean it up a bit and then issue a pull request08:44
alesis-novikif it has plotting, I should put it in /examples/undocumented/python_modular/graphical/  ?08:45
@sonney2kalesis-novik, yes - but make sure you also do some non-graphical variant for illustration that can be run automagically for regression tests08:46
-!- alesis-novik [~alesis@188.74.87.206] has quit [Quit: Leaving]09:16
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun10:45
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]10:49
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun11:01
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun11:15
@sonney2kblackburn, I have a new radical idea11:28
@sonney2khow about renaming the non-swig interfaces into *-legacy?11:28
blackburnpython -> python_legacy?11:28
@sonney2kyeah11:28
blackburnwhy not :)11:29
blackburnis it so radical?11:29
@sonney2kand renaming the modular ones to just python11:29
blackburnahhh11:29
@sonney2kblackburn, I mean currently everyone thinks that python is the thing to install11:43
@sonney2kthough the modular one is what people should use11:43
@sonney2ksame for other languages11:44
@sonney2kof course we still have issues with octave / swig...11:44
@sonney2kand with R too11:44
-!- heiko [~heiko@134.91.10.200] has joined #shogun11:58
-!- blackburn [~blackburn@188.122.238.99] has quit [Ping timeout: 255 seconds]12:05
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun12:05
blackburnyeah, I understand12:06
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun12:48
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds]12:51
-!- in3xes__ [~in3xes@180.149.49.227] has joined #shogun12:51
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds]12:52
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun13:01
-!- in3xes__ [~in3xes@180.149.49.227] has quit [Ping timeout: 260 seconds]13:04
-!- in3xes_ is now known as in3xes13:11
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun13:21
-!- in3xes__ [~in3xes@180.149.49.227] has joined #shogun13:21
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 246 seconds]13:25
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Ping timeout: 276 seconds]13:25
-!- in3xes__ [~in3xes@180.149.49.227] has quit [Quit: Leaving]13:27
-!- in3xes1 [~in3xes@210.212.58.111] has joined #shogun13:38
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 258 seconds]13:42
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun15:01
-!- in3xes1 [~in3xes@210.212.58.111] has quit [Ping timeout: 258 seconds]15:04
@sonney2kheiko, how is it going with your bug hunt?15:37
@sonney2kheiko seems to be in some deep deep forest....16:08
blackburnyeah just like me16:09
blackburnI finally managed to get some example where arpack shift-inverse mode fails16:10
blackburnI guess my fault, not these arpack guys :D16:10
@sonney2kblackburn, me too btw... I have some numerical problem where the order of summation matters a lot :D16:10
blackburnnumerical issues all around the world16:11
-!- in3xes_ is now known as in3xes16:11
heikosonney2k, hey, i was having a break :)16:54
heikobug hunt making progress. only one method remaining, but I am writing some more in-depth test cases now to prevent this something like this happening again :)16:55
@sonney2kheiko, the worst excuse I ever heard :D16:55
@sonney2kweitermachen then :)16:56
heikoi am dabei :)16:56
heikooh its working already16:57
heikoonly a memory leak16:57
heikoabout 200 frees of 150016:57
heikosome loop or so16:57
heikosonney2k, a one armed blind man just stole my bugfixes ;)17:00
blackburnsonney2k: putin made a bug in shift-invert mode of arpack17:01
blackburnheiko: try this http://make-everything-ok.com/17:03
heikolol, :)17:06
heikommh i have to check my pereption ...17:07
heikooh well, found another leak :)17:07
bettybooho heiko17:07
blackburnoh okay my solution is to change 'LM' to 'LA'17:35
blackburn:D17:35
blackburnthe only sym to change everything17:36
-!- heiko [~heiko@134.91.10.200] has quit [Ping timeout: 258 seconds]18:10
CIA-32shogun: Sergey Lisitsyn master * rd66238c / src/libshogun/preprocessor/LocallyLinearEmbedding.cpp : Changed arpack mode at Locally Linear Embedding - http://bit.ly/qcWeAy18:34
CIA-32shogun: Sergey Lisitsyn master * r259c5d3 / src/libshogun/lib/arpack.cpp : Improved arpack wrapper - http://bit.ly/p3cqSY18:34
-!- f-x [~user@117.192.215.88] has joined #shogun18:37
-!- heiko [~heiko@134.91.10.200] has joined #shogun18:49
heikofinally!!!!19:06
heikoold examples work again19:06
heikooh no, one more example leaks :) hopefully last one19:15
heikopull request sent :)19:35
blackburnheiko: i'll push it in a min19:37
CIA-32shogun: Sergey Lisitsyn master * r2f7d4e0 / (11 files in 3 dirs): Merge branch 'master' of git://github.com/karlnapf/shogun (+6 more commits...) - http://bit.ly/paFvmR19:38
heikoblackburn, thanks :)19:38
CIA-32shogun: Sergey Lisitsyn master * ra0c02ba / (5 files): Merge branch 'master' of https://github.com/sploving/shogun - http://bit.ly/qqnIqF19:39
@sonney2kheiko, well done...20:36
heikosoney2k, thanks :)20:36
bettyboocool20:36
@sonney2kI am still fighting with the gamma function20:36
heikowhat does it (not) do?20:38
heikoyes, simple example runs :)20:38
heikook, gotta go20:39
heikoits late and I have been sitting here for all day, time for some non-computer activity :)20:39
bettyboorotfl20:39
heikosonney2k, have a nice evening20:39
@sonney2kheiko, I have a deviation of 1e-14 here and don't get it :(20:44
-!- heiko [~heiko@134.91.10.200] has quit [Ping timeout: 258 seconds]20:45
@sonney2kblackburn, looks like heiko didn't do a pull --rebase21:27
blackburnsonney2k: where?21:27
@sonney2kall patches in his new pull request are on top21:28
@sonney2khttps://github.com/shogun-toolbox/shogun/pull/192/files21:28
blackburnah yes21:28
-!- f-x [~user@117.192.215.88] has quit [Ping timeout: 260 seconds]21:29
-!- mlsec [~mlsec@bane.ml.tu-berlin.de] has quit [Read error: Operation timed out]22:37
-!- mlsec [~mlsec@bane.ml.tu-berlin.de] has joined #shogun22:38
-!- mode/#shogun [+o mlsec] by ChanServ22:38
blackburnsonney2k: brand new developer ;)23:15
@sonney2kblackburn, you recruited him?23:16
blackburnsonney2k: yeap23:17
@sonney2k:)23:17
@sonney2kthen do your merging job :)23:17
blackburnsonney2k: of course, we are now doing some pull request training ;)23:17
@sonney2kvery good23:17
CIA-32shogun: Evgeniy Andreev master * rc4fbb4a / (3 files): Example preprocessor_isomap_modular is splitted into 2 parts with ClassicIsomap and LandmarkIsomap - http://bit.ly/p4YOdm23:21
CIA-32shogun: Evgeniy Andreev master * r1b2802d / examples/undocumented/python_modular/preprocessor_landmarkisomap_modular.py : Modify - LandmarkIsomap example - http://bit.ly/o1Zkj523:21
blackburnsonney2k: here it is23:22
blackburn;)23:22
@sonney2kbah, I am giving up on the gamma function23:30
@sonney2kI have to do some more enjoyable programming tomorrow23:30
@sonney2kotherwise these gamma rays will kill me23:30
blackburnwhat's up with your gamma function?23:31
@sonney2kblackburn, I have a deviation of 1e-14 with lgamma and what I get from matlab23:32
@sonney2kthis doesn't sound bad but this deviation changes the result from 0.82 -> 0.7 in the end23:32
blackburnhuh23:32
@sonney2kblackburn, even lgammal (for long double) didn't help23:33
@sonney2kmaybe I have been staring at this for too long now though23:33
blackburnlgamma looks like something you could believe..23:33
blackburnbut matlab too :D23:33
blackburnsonney2k: try http://www.wolframalpha.com/ :D23:34
@sonney2kblackburn, I did :)23:34
blackburnand?23:34
@sonney2kI gave up before checking23:34
@sonney2kwill try tomorrow23:35
@sonney2k(I figured out that lgamma is the problem after 1 day of debugging...)23:35
blackburnhuh23:35
blackburnwhere do you use gamma function? :)23:36
@sonney2kin this p value computation stuff23:37
CIA-32shogun: Soeren Sonnenburg master * r8e6b8e0 / (2 files):23:39
CIA-32shogun: - introduce lgammal23:39
CIA-32shogun: - add display_vector for long double23:39
CIA-32shogun: - change p-value computation to use floatmax_t and logarithmic_sum to increase robustness - http://bit.ly/nmr6u923:39
CIA-32shogun: Soeren Sonnenburg master * rfbeb540 / (21 files in 6 dirs): Merge branch 'master' of github.com:shogun-toolbox/shogun - http://bit.ly/oyV8du23:39
CIA-32shogun: Soeren Sonnenburg master * r8b3c5d3 / src/csharp_modular/swig_typemaps.i : remove obsolete typemaps from csharp - nothing in cs works yet though - http://bit.ly/nOAnkS23:43
@sonney2kok, giving up for today...23:43
@sonney2kcu blackburn !23:43
blackburnsonney2k: see ya23:43
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun23:49
--- Log closed Thu Jul 14 00:00:18 2011
« Tuesday, 2011-07-12 Index Thursday, 2011-07-14 »

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!