--- Log opened Fri Aug 01 00:00:39 2014 | ||
-!- lambday [7aa773d8@gateway/web/freenode/ip.122.167.115.216] has quit [Ping timeout: 246 seconds] | 00:14 | |
-!- soumyaC [uid15286@gateway/web/irccloud.com/x-yzsfnwcxhczbufpc] has quit [Quit: Connection closed for inactivity] | 04:16 | |
-!- travis-ci [~travis-ci@ec2-54-204-222-14.compute-1.amazonaws.com] has joined #shogun | 04:49 | |
travis-ci | [travis-ci] it's khalednasr's turn to pay the next round of drinks for the massacre he caused in khalednasr/shogun: http://travis-ci.org/khalednasr/shogun/builds/31391267 | 04:49 |
---|---|---|
-!- travis-ci [~travis-ci@ec2-54-204-222-14.compute-1.amazonaws.com] has left #shogun [] | 04:49 | |
-!- travis-ci [~travis-ci@ec2-50-17-35-184.compute-1.amazonaws.com] has joined #shogun | 05:22 | |
travis-ci | [travis-ci] it's khalednasr's turn to pay the next round of drinks for the massacre he caused in khalednasr/shogun: http://travis-ci.org/khalednasr/shogun/builds/31391311 | 05:22 |
-!- travis-ci [~travis-ci@ec2-50-17-35-184.compute-1.amazonaws.com] has left #shogun [] | 05:22 | |
-!- sonney2k [~shogun@7nn.de] has joined #shogun | 05:41 | |
-!- Netsplit *.net <-> *.split quits: sonney2k_ | 05:47 | |
-!- lambday [c40f1064@gateway/web/freenode/ip.196.15.16.100] has joined #shogun | 06:44 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 08:08 | |
shogun-notifier- | shogun: Abinash Panda :develop * abadcc8 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/abadcc8b899750efb9f78b31a6d78253d792f752 | 08:08 |
shogun-notifier- | shogun: Added HierarchicalMultilabelModel class | 08:08 |
shogun-notifier- | shogun: Abinash Panda :develop * 377ad64 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/377ad648da57259df4a33ea760c50e9116f4a240 | 08:08 |
shogun-notifier- | shogun: Modified HierarchicalMultilabelModel: | 08:08 |
shogun-notifier- | shogun: * Changed get_attr_vector -> get_label_vector | 08:08 |
shogun-notifier- | shogun: * Minor refactorings in tests and source code | 08:08 |
shogun-notifier- | shogun: tklein23 :develop * adda76f / / (3 files): https://github.com/shogun-toolbox/shogun/commit/adda76f5bc96e6e8006c1a74bd7785a0cc543e7d | 08:08 |
shogun-notifier- | shogun: Merge pull request #2426 from abinashpanda/develop | 08:08 |
shogun-notifier- | shogun: | 08:08 |
shogun-notifier- | shogun: Added HierarchicalMultilabelModel class | 08:08 |
shogun-buildbot | build #2453 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2453 blamelist: Abinash Panda <abinash.panda.ece10@itbhu.ac.in> | 08:21 |
shogun-buildbot | build #2454 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2454 blamelist: Abinash Panda <abinash.panda.ece10@itbhu.ac.in>, tklein23 <tklein23@users.noreply.github.com> | 08:34 |
shogun-buildbot | build #35 of osx2 - modular_interfaces is complete: Failure [failed csharp modular] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20modular_interfaces/builds/35 blamelist: Abinash Panda <abinash.panda.ece10@itbhu.ac.in> | 09:09 |
shogun-buildbot | build #36 of osx2 - modular_interfaces is complete: Failure [failed csharp modular] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20modular_interfaces/builds/36 blamelist: Abinash Panda <abinash.panda.ece10@itbhu.ac.in>, tklein23 <tklein23@users.noreply.github.com> | 09:13 |
-!- lambday [c40f1064@gateway/web/freenode/ip.196.15.16.100] has quit [Ping timeout: 246 seconds] | 09:36 | |
-!- lambday [c40f1069@gateway/web/freenode/ip.196.15.16.105] has joined #shogun | 10:45 | |
-!- HeikoS [~heiko@nat-190-6.internal.eduroam.ucl.ac.uk] has joined #shogun | 11:08 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:08 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 11:08 | |
@besser82 | HeikoS, G'Morning ^^ Back in London? | 11:09 |
@HeikoS | besser82: hi, yes :) | 11:09 |
@besser82 | HeikoS, journey was good? | 11:09 |
@besser82 | You saw my email about the torrents? | 11:10 |
@besser82 | so if you want the 'raw'-stuff just go on and leech it ;) | 11:11 |
@besser82 | HeikoS, ^^ | 11:11 |
@HeikoS | besser82: hey | 11:15 |
@HeikoS | besser82: ok will do, just one file, i wondered about the sound | 11:15 |
@besser82 | HeikoS, allrighty... ^^ | 11:15 |
@HeikoS | besser82: sorry for delayed response | 11:15 |
@besser82 | HeikoS, np ^^ in Trasmission you can choose the file from the torrent you want to load, then ^^ | 11:16 |
@besser82 | HeikoS, take the raw-video thing then... | 11:16 |
@HeikoS | besser82: ok thanks, checking | 11:16 |
@besser82 | HeikoS, np yw ^^ | 11:17 |
@HeikoS | besser82: downloading | 11:21 |
@HeikoS | besser82: thanks so much for putting the videos up so promtly, amazing! | 11:21 |
@besser82 | HeikoS, many thanks!!! =) | 11:22 |
@besser82 | HeikoS, I can see no torrent connection from you, yet... | 11:23 |
@HeikoS | mmh | 11:24 |
@HeikoS | i might not be allowed to torrent at work | 11:24 |
@besser82 | ic | 11:24 |
@besser82 | so just start downloading somewhere you are allowed to; I'll keep seeding... | 11:24 |
-!- HeikoS [~heiko@nat-190-6.internal.eduroam.ucl.ac.uk] has quit [Read error: Connection reset by peer] | 12:12 | |
-!- HeikoS [~heiko@nat-190-6.internal.eduroam.ucl.ac.uk] has joined #shogun | 12:12 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:12 | |
-!- HeikoS [~heiko@nat-190-6.internal.eduroam.ucl.ac.uk] has quit [Client Quit] | 12:15 | |
-!- HeikoS [~heiko@nat-190-6.internal.eduroam.ucl.ac.uk] has joined #shogun | 12:37 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:37 | |
-!- lambday [c40f1069@gateway/web/freenode/ip.196.15.16.105] has quit [Quit: Page closed] | 12:39 | |
-!- HeikoS [~heiko@nat-190-6.internal.eduroam.ucl.ac.uk] has quit [Client Quit] | 12:39 | |
-!- thoralf [~thoralf@91-64-98-140-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 12:43 | |
@besser82 | I'm currently creating new chroots for fc21, fc_rawhide and CentOS7. | 13:16 |
@besser82 | Will upload them in a few minutes, so you can get them running on fatbot. Links are following... | 13:17 |
@besser82 | wiking, ^^ | 13:17 |
@besser82 | wiking, upload is running... you can fetch them from here: https://besser82.fedorapeople.org/chroots/ | 13:21 |
-!- kislay [~Abhijeet@220.227.97.99] has joined #shogun | 13:53 | |
sonne|work | wiking: seen my email? | 14:38 |
-!- kislay [~Abhijeet@220.227.97.99] has quit [Read error: Connection reset by peer] | 14:45 | |
-!- lambday [7aa773d8@gateway/web/freenode/ip.122.167.115.216] has joined #shogun | 15:52 | |
-!- HeikoS [~heiko@nat-190-138.internal.eduroam.ucl.ac.uk] has joined #shogun | 16:07 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:07 | |
lambday | HeikoS: hi! | 16:16 |
@HeikoS | lambday: hi | 16:17 |
lambday | HeikoS: I was checking the approach you suggested - one issue is.. we are not computing trace(inv(...)) but trace(A*inv(B)*C*inv(D)) sort of thing... | 16:18 |
lambday | the code I pasted computes this A*inv(B) without storing inv(B) | 16:18 |
lambday | and then for trace we can use the trick you mentioned | 16:19 |
@HeikoS | lambday: ok then | 16:20 |
@HeikoS | lambday: just wanted to make sure we never store the inv :) | 16:20 |
lambday | HeikoS: just out of curiosity - what's so bad about storing the inverse? except the extra memory requirement | 16:20 |
@HeikoS | lambday: its unstable to do this | 16:20 |
@HeikoS | lambday: might easily explode | 16:20 |
@HeikoS | lambday: and one gets lots of rounding errors | 16:21 |
lambday | HeikoS: okay | 16:21 |
@HeikoS | lambday: matlab these days even gives a warning when one does inv(A)*x | 16:21 |
@HeikoS | and tell you to do "\" operator | 16:21 |
lambday | HeikoS: alright | 16:21 |
lambday | HeikoS: so do you think this approach is okay? https://gist.github.com/lambday/66e637b46109e132181d#file-inverse_benchmark-cpp-L74 | 16:22 |
@HeikoS | lambday: for matrices that are a bit largetr and a bit less good conditioned, things fall apart very quickly | 16:22 |
@HeikoS | lambday: solving the matrix times identity vector thing? | 16:22 |
@HeikoS | lambday: yeah thats better, but even better would to avoid all this business at all and do svd maybe | 16:23 |
lambday | HeikoS: yes... and then using the soln vectors to compute matrix multiplication using dot on the fly | 16:23 |
lambday | we already have dot in linalg | 16:23 |
@HeikoS | lambday: yeah but thats also unstable | 16:23 |
lambday | so that's good | 16:23 |
@HeikoS | lambday: so if we have A*inv(B)*C*inv(D) | 16:23 |
@HeikoS | lambday: that can be computed and stored | 16:24 |
lambday | HeikoS: the whole thing? we actually need the trace of this | 16:24 |
@HeikoS | lambday: yeah sorry | 16:24 |
@HeikoS | lambday: too many things going on at once ;) | 16:25 |
@HeikoS | lambday: no i think your approach is fine then | 16:25 |
@HeikoS | lambday: should be ok | 16:25 |
lambday | HeikoS: hehe.. okay let me know when you're free | 16:25 |
@HeikoS | lambday: but maybe test it a bit | 16:25 |
@HeikoS | lambday: free now, thats fine, just giving excuses why i dont know things exactly | 16:26 |
@HeikoS | so yeah, do this one, | 16:26 |
@HeikoS | with the dot in linalg, that is axtually great | 16:26 |
@HeikoS | lambday: the loop should be in poarallel | 16:26 |
@HeikoS | is the chol.solve thread safe? | 16:26 |
lambday | HeikoS: yeah that too I did in this code... gained a slight improvement | 16:26 |
@HeikoS | cool | 16:26 |
lambday | oh that... umm not sure | 16:27 |
lambday | I was talking about computing a whole bunch of dots in parallel | 16:27 |
@HeikoS | lambday: ok that is also fine | 16:27 |
@HeikoS | lambday: i mean you can move the system solve inside there (more expensive than dot in fact) | 16:27 |
@HeikoS | so a few (chol.solve, dot-product) things run in parallel | 16:27 |
@HeikoS | lambday: just saw you have an omp pragma there | 16:28 |
@HeikoS | lambday: does this work if you move it up a line? | 16:28 |
lambday | so inside each thread - solve one system for each col of identity matrix and then compute dots | 16:28 |
@HeikoS | lambday: in fact | 16:28 |
@HeikoS | lambday: its good as it is | 16:28 |
@HeikoS | lambday: so keep it :) | 16:28 |
@HeikoS | lambday: i gotta go to a talk now, let me know how things go | 16:29 |
lambday | HeikoS: alright | 16:29 |
@HeikoS | lambday: very nice that you investigated this properly! | 16:29 |
lambday | HeikoS: as of now I'm just using eigen3 cholesky and dot - will change it when linalg is there in develop | 16:29 |
-!- yorkerlin [81617cd0@gateway/web/freenode/ip.129.97.124.208] has joined #shogun | 16:34 | |
-!- PirosB3 [~pirosb3@host49-195-dynamic.21-87-r.retail.telecomitalia.it] has joined #shogun | 16:41 | |
-!- yorkerlin [81617cd0@gateway/web/freenode/ip.129.97.124.208] has quit [Ping timeout: 246 seconds] | 17:46 | |
-!- lambday [7aa773d8@gateway/web/freenode/ip.122.167.115.216] has quit [Ping timeout: 246 seconds] | 18:14 | |
-!- iglesiasg [524b8e0b@gateway/web/freenode/ip.82.75.142.11] has joined #shogun | 19:32 | |
-!- kislay [~Abhijeet@210.212.61.185] has joined #shogun | 19:34 | |
-!- iglesiasg is now known as iglesiasg_ | 19:36 | |
@lisitsyn | HeikoS: hey | 19:38 |
-!- iglesiasg [~iglesias@524B8E0B.cm-4-4c.dynamic.ziggo.nl] has joined #shogun | 19:39 | |
@lisitsyn | another try | 19:40 |
@lisitsyn | iglesiasg hey :) | 19:40 |
-!- iglesiasg [~iglesias@524B8E0B.cm-4-4c.dynamic.ziggo.nl] has left #shogun [] | 19:40 | |
iglesiasg_ | lisitsyn: hello hello | 19:43 |
@lisitsyn | iglesiasg_: share your experience on the workshop ;) | 19:43 |
@lisitsyn | is there anything new going on? | 19:43 |
@lisitsyn | I feel lost :) | 19:43 |
iglesiasg_ | it was pretty cool :-) | 19:43 |
-!- iglesiasg_ is now known as iglesiasg | 19:44 | |
iglesiasg | lisitsyn: why do you feel lost? | 19:45 |
@lisitsyn | iglesiasg: well I have no idea what you guys discussed there :) | 19:45 |
@lisitsyn | so I am a bit out of the process | 19:45 |
iglesiasg | apart from the stuff of the videos | 19:46 |
iglesiasg | we discussed a bit on the mentor summit | 19:46 |
iglesiasg | I must buy by plane tickets asap btw | 19:46 |
@lisitsyn | didn't you yet? | 19:46 |
iglesiasg | well, no | 19:47 |
iglesiasg | it is more than two and a half months left | 19:47 |
iglesiasg | but yes, I must do it soon anyway | 19:47 |
@lisitsyn | ha true | 19:47 |
@lisitsyn | I just thought they are messing with these returns already | 19:47 |
@lisitsyn | POs whatever | 19:48 |
@lisitsyn | iglesiasg: I am curious whether anything about kaggle is still of interest for any of us | 19:49 |
@lisitsyn | it died immediately :D | 19:49 |
iglesiasg | lisitsyn: we didn't bring up that topic in fact | 19:49 |
@lisitsyn | iglesiasg: I see | 19:51 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 20:13 | |
shogun-notifier- | shogun-web: Fernando Iglesias :master * e230c3f / static/workshop/2014/ (7 files): https://github.com/shogun-toolbox/shogun-web/commit/e230c3f75a668a906e682999d1fd344246f9c65c | 20:13 |
shogun-notifier- | shogun-web: Slides workshop 2014. | 20:13 |
-!- HeikoS [~heiko@nat-190-138.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 20:22 | |
-!- iglesiasg [524b8e0b@gateway/web/freenode/ip.82.75.142.11] has quit [Quit: Page closed] | 20:34 | |
-!- mE\Ta [~rain@95.211.184.195] has quit [Ping timeout: 256 seconds] | 21:59 | |
-!- kislay [~Abhijeet@210.212.61.185] has quit [Read error: Connection reset by peer] | 22:00 | |
shogun-notifier- | shogun: Fernando Iglesias :develop * 45fd6db / doc/ipython-notebooks/statistics/mmd_two_sample_testing.ipynb: https://github.com/shogun-toolbox/shogun/commit/45fd6dbb1abbf3336e0384acf42b9fb05fc3f899 | 22:15 |
shogun-notifier- | shogun: Minor fixes to mmd notebook. | 22:15 |
shogun-buildbot | build #2455 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2455 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 22:23 |
-!- mE\Ta [~rain@95.211.184.195] has joined #shogun | 22:28 | |
shogun-buildbot | build #37 of osx2 - modular_interfaces is complete: Failure [failed csharp modular] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20modular_interfaces/builds/37 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 22:43 |
-!- pickle27 [~pickle27@192-0-136-118.cpe.teksavvy.com] has joined #shogun | 23:45 | |
--- Log closed Sat Aug 02 00:00:00 2014 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!