Open in new window / Try shogun cloud
--- Log opened Mon Mar 21 00:00:05 2016
--- Day changed Mon Mar 21 2016
@HeikoSlambday: yes definitely00:00
@HeikoSlambday: but also the fact that it is easy to extend linalg00:00
@lambdaythe 'custom(...)' thing is a good option for that00:00
@lambdaygood that we discussed it man00:00
@HeikoSand you smashed it with the prototype00:00
@HeikoSso cool00:00
@HeikoSIt feels like this is how it should be00:01
@HeikoSso hopefully we can use this stuff soon00:01
@HeikoSI think we should aim to put the GP framework on the back of it first00:01
@HeikoSsince it is new, we dont really break compatibility00:01
@HeikoS(because of c++11)00:01
@HeikoSso a good test field00:02
@HeikoSas well as all the NN stuff00:02
@lambdayHeikoS: if I get to work on that a bit, maybe I'll learn a bit of GP :D00:02
@HeikoSthese two should use the linalg first00:02
@HeikoSif that goes well, we can use it everywhere00:02
@HeikoSyorkerlin is very keen to put GPs on the GPU00:02
@lambdayyeah.. that sounds good00:02
@HeikoSlisitsyn: how's swig?00:06
lisitsynHeikoS: no idea why it fails again :)00:07
@HeikoSok, thats ALL shogun emails answered00:07
lisitsynHeikoS: thanks a lot for handling this00:07
@HeikoSwe got some good stuff going on with the benchmarks00:07
@HeikoSlisitsyn: I hope that someone joins me at some point00:08
@HeikoSlisitsyn: btw about the sampling00:09
@HeikoSthink about algorithms in terms of "how good per unit time"00:09
@HeikoSyou can think of a plot00:09
lisitsynHeikoS: mcmc needs quite a lot of iterations in my experience00:10
lisitsyneven for simple models00:10
@HeikoSdepends on what you mean by iterations00:10
@HeikoS a single mcmc iteration has to touch all data00:11
@HeikoSthat is expensive00:11
@HeikoSbut good samplers converge fast if used correctly00:11
@HeikoSbut anyways00:11
@HeikoSthe point is:00:11
@HeikoSmcmc gets arbritarily good00:11
@HeikoSthat is if you continue to run it, it gets better00:11
@HeikoSvariational does not00:11
@HeikoSthats the main difference00:11
@HeikoSlisitsyn, lambday see you tomorrow!00:12
lisitsynI see00:12
lisitsynsee you00:12
@lambdayHeikoS: see you!00:12
@lambdayme too leaving.. it's 4:40 AM here00:12
-!- HeikoS [] has quit [Quit: Leaving.]00:12
-!- travis-ci [] has joined #shogun00:25
travis-ciit's Soumyajit De's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun:
-!- travis-ci [] has left #shogun []00:25
@lambdaylisitsyn: any idea on how to support namespace structure for class_list ?00:37
lisitsynlambday: haven't checked yt00:37
@lambdayhmm me neither00:38
@lambdaythis should be a useful thing to support00:38
@lambdaybut for interfaces, how this is gonna work?00:38
@lambdayfrom modshogun import subnamespace - does it work that way?00:39
shogun-notifier-shogun: youssef :feature/bigtest * cea770b / src/shogun/multiclass/QDA.cpp:
shogun-notifier-shogun: remove rotations transpose01:09
shogun-notifier-shogun: lambday :feature/bigtest * cc68281 / src/shogun/statistics/experimental/internals/ (5 files):
shogun-notifier-shogun: documentation and code-style changes01:09
shogun-notifier-shogun: lambday :feature/bigtest * fd95662 / src/shogun/statistics/experimental/ (3 files):
shogun-notifier-shogun: added hypothesis test (experimental) base class01:09
shogun-buildbotbuild #3671 of deb1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: lambday <>, youssef <>01:11
@lambdaygood night lisitsyn01:13
-!- lambday [6a33f351@gateway/web/freenode/ip.] has quit []01:13
-!- travis-ci [] has joined #shogun01:39
travis-ciit's lambday's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun:
-!- travis-ci [] has left #shogun []01:39
-!- GandalfTheWizard [~ian@] has quit [Ping timeout: 240 seconds]02:37
-!- GandalfTheWizard [~ian@] has joined #shogun02:39
-!- GandalfTheWizard [~ian@] has quit [Ping timeout: 260 seconds]02:46
-!- GandalfTheWizard [~ian@] has joined #shogun02:49
shogun-buildbotbuild #975 of nightly_all is complete: Failure [failed test]  Build details are at  blamelist: Heiko Strathmann <>, youssef <>03:06
-!- GandalfTheWizard [~ian@] has quit [Ping timeout: 246 seconds]04:04
-!- GandalfTheWizard [~ian@] has joined #shogun04:08
shogun-buildbotbuild #1102 of nightly_default is complete: Failure [failed test notebooks]  Build details are at  blamelist: Heiko Strathmann <>, youssef <>04:09
-!- shogun-notifier- [] has quit [Quit: transmission timeout]04:09
-!- Saurabh7 [~Saurabh7@] has joined #shogun05:39
-!- Saurabh7 [~Saurabh7@] has quit [Ping timeout: 246 seconds]06:09
-!- Saurabh7 [~Saurabh7@] has joined #shogun06:23
-!- amod95 [uid67143@gateway/web/] has joined #shogun07:30
-!- sonne|work [] has quit [Quit: Leaving.]09:14
-!- besser82 [~besser82@fedora/besser82] has joined #shogun09:27
-!- mode/#shogun [+o besser82] by ChanServ09:27
-!- Saurabh7 [~Saurabh7@] has quit [Quit: Leaving]09:30
-!- amod95 [uid67143@gateway/web/] has quit [Quit: Connection closed for inactivity]09:48
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 240 seconds]11:11
-!- sonne|work [] has joined #shogun11:40
-!- HeikoS [] has joined #shogun12:55
-!- mode/#shogun [+o HeikoS] by ChanServ12:55
-!- shogun-notifier- [] has joined #shogun12:59
shogun-notifier-shogun-data: Sanuj :master * 4e01b47 / testsuite/meta/regression/kernel_ridge_regression.dat:
shogun-notifier-shogun-data: update krr integration test data12:59
shogun-notifier-shogun-data: Heiko Strathmann :master * 77cba34 / testsuite/meta/regression/kernel_ridge_regression.dat:
shogun-notifier-shogun-data: Merge pull request #85 from sanuj/master12:59
shogun-notifier-shogun-data: update krr integration test data12:59
-!- Nebula_ [~chatzilla@] has joined #shogun13:03
-!- Nebula_ [~chatzilla@] has left #shogun []13:05
-!- HeikoS [] has quit [Ping timeout: 248 seconds]13:06
-!- besser82 [~besser82@fedora/besser82] has joined #shogun13:10
-!- mode/#shogun [+o besser82] by ChanServ13:10
-!- GandalfTheWizar1 [~ian@] has joined #shogun14:41
-!- GandalfTheWizard [~ian@] has quit [Ping timeout: 260 seconds]14:41
-!- arasu [daf8068c@gateway/web/freenode/ip.] has joined #shogun15:21
-!- shubham____ [uid57367@gateway/web/] has joined #shogun15:25
-!- arasu [daf8068c@gateway/web/freenode/ip.] has quit [Quit: Page closed]15:26
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 264 seconds]15:31
-!- sanuj [0e8bc402@gateway/web/freenode/ip.] has joined #shogun15:36
-!- arasu [daf8068c@gateway/web/freenode/ip.] has joined #shogun15:36
-!- arasu [daf8068c@gateway/web/freenode/ip.] has quit [Client Quit]15:40
-!- shogun-notifier- [] has quit [Quit: transmission timeout]15:59
-!- arasu [daf8068c@gateway/web/freenode/ip.] has joined #shogun16:34
-!- arasu_ [daf8068c@gateway/web/freenode/ip.] has joined #shogun16:45
-!- arasu [daf8068c@gateway/web/freenode/ip.] has quit [Ping timeout: 252 seconds]16:47
-!- arasu_ [daf8068c@gateway/web/freenode/ip.] has quit [Ping timeout: 252 seconds]16:54
-!- HeikoS [] has joined #shogun16:59
-!- mode/#shogun [+o HeikoS] by ChanServ16:59
-!- livus [~suvil@] has joined #shogun17:02
-!- sanuj [0e8bc402@gateway/web/freenode/ip.] has quit [Ping timeout: 252 seconds]17:04
-!- besser82 [~besser82@fedora/besser82] has joined #shogun17:08
-!- mode/#shogun [+o besser82] by ChanServ17:08
-!- shogun-notifier- [] has joined #shogun17:10
shogun-notifier-shogun: Xbar :develop * 0b4f8d6 / / (3 files):
shogun-notifier-shogun: Bugfix: 307517:10
shogun-notifier-shogun: Fix issue# 307517:10
shogun-notifier-shogun: Documented max_iteration17:10
shogun-notifier-shogun: Heiko Strathmann :develop * 40228bc / / (3 files):
shogun-notifier-shogun: Merge pull request #3097 from Xbar/develop17:10
shogun-notifier-shogun: Bugfix: 307517:10
shogun-buildbotbuild #3672 of deb1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Heiko Strathmann <>, Xbar <>17:12
-!- arasu [daf8068c@gateway/web/freenode/ip.] has joined #shogun17:17
-!- shubham____ [uid57367@gateway/web/] has quit [Quit: Connection closed for inactivity]17:27
-!- arasu [daf8068c@gateway/web/freenode/ip.] has quit [Quit: Page closed]17:37
-!- mizari [] has joined #shogun17:41
-!- lambday [6a33e882@gateway/web/freenode/ip.] has joined #shogun17:43
-!- mode/#shogun [+o lambday] by ChanServ17:43
@HeikoSlambday: we could maybe have a subfolder for hypothesis testing, rather than statistics17:44
@lambdayHeikoS: yeah that would be good17:45
@HeikoSlambday: hypothesis_testing17:46
@lambdayHeikoS: but what could be a good name for that? shogun/statistics/hypothesistest is quite a mouthful :/17:46
@HeikoSrename statistics17:46
@lambdayyeah that makes sense17:46
@HeikoSon the other hand, statistics also contains feature selection code doesnt it?17:46
@lambdaynah that's in preprocessor17:47
@HeikoSbut all these are dependence measure in some or another way17:47
@HeikoSah ok17:47
@HeikoSwell then17:47
@HeikoSrename statistics (pretty useless name anyways ;) )17:47
@lambdayHeikoS: cooking dinner now :D will start in a while17:48
@HeikoSlambday: what are you cooking?17:48
@lambdaybtw does arthur follow cricket? he said something regarding the australian cricket team - today aussies are playing against bangladesh17:48
@HeikoSlambday: eeehm, no idea ;)17:48
@HeikoShe travels to Japan today so I guess he will just watch zombie movies in the plane17:49
@lambdayHeikoS: spinach - popeye the sailor man17:49
@HeikoShaha :D17:49
@HeikoSwe got some nice south indian places here in London, I am really ucrious how you will like them17:49
@HeikoSbtw, any news on this visa thing, or just waiting for time to pass?17:49
@lambdayHeikoS: hehe I am not a big fan of south indian dishes in general..17:50
@HeikoSreally?, interesting17:50
@lambdayHeikoS: yeah I cc-ed you the conversation I had with reign.. I gotta wait17:50
@lambdayno other way17:50
* lambday brb17:56
-!- sanuj [0e8bc402@gateway/web/freenode/ip.] has joined #shogun17:58
sanujHeikoS: there?18:00
@HeikoSsanuj: yes, hi!18:00
sanujHeikoS: hello18:00
sanujHeikoS: we dont need a setter in train method right?18:01
@HeikoSwhat do you mean by that?18:01
sanujllt will update the preallocated alpha vector18:01
@HeikoSsanuj: ah yes, you pass the preallocated vector to the new method, which will update it18:02
sanujHeikoS: like 'eigen_y = llt.solve(eigen_y);'18:02
sanujso no need to set alpha again in train18:02
@HeikoSbut dont do the above18:02
@HeikoSyou will overwrite the labels18:02
@HeikoSeigen_alpha = llt.solve(eigen_y)18:02
sanujHeikoS: yeah18:02
sanuji'll take care of it18:03
@HeikoScool, looking forward18:03
sanujHeikoS: so the solve_krr_system() will return void18:03
sanujor maybe a bool18:03
@HeikoSbool is good, yes18:04
@HeikoSsanuj: point is to make it easy to overload this method18:05
@HeikoSfor the approximate versions18:05
@HeikoSwhich are based on a different kernel matri18:06
sanujHeikoS: yes18:07
@lambdaywhat's ShoGaY18:13
@lambdaywhois ShoGaY18:13
@lambdayShoGaY: hello18:14
sanujlambday: i think it's a bot18:16
sanujnot sure18:16
sanujask wiking18:16
@lambdayI too think so18:16
@lambdaythe name is funny haha18:17
@lambdaywhy your name isn't coming on tab18:17
@HeikoStroll bot ;)18:18
@lambdaynah no luck18:19
@lambdayShoGaY: commands18:19
@lambdayShoGaY: dance18:19
@lambdayShoGaY: boo!18:19
@lambdayshogun-buildbot: dance18:19
@HeikoSdoes this work for you?18:19
@HeikoSshogun-buildbot: boo!18:19
shogun-buildbotWhat you say!18:19
@lambdayshogun-buildbot: What you say!18:20
shogun-buildbotWhat you say!18:20
@lambdayHeikoS: was the link for me or sanuj18:21
@lambdaybangladesh making the aussies fight! some sight to behold!18:23
@HeikoSlambday: github was down for a second18:23
@HeikoSbut now its back up18:23
@HeikoSso forget about the link18:23
sanujHeikoS: is that for your phd18:23
@HeikoSsanuj: kind of yes18:23
@HeikoScrappy python code is what you do in phds ;)18:24
sanujis it about to get over?18:24
@HeikoS1.5 more years18:24
sanujgood luck :)18:25
@lambdayHeikoS: in UCL you're allowed to take another job while you're at PhD right?18:25
sanujlambday: how will you find the time?18:25
@lambdaysanuj: HeikoS does18:25
@lambdayI guess18:25
sanujshogun + phd + another job18:25
@lambdayit's really useful to have industry exposure IMO18:26
@lambdaysanuj: he's superman!18:26
@lambdaykarlnapf is short for clark kent, in case you didn't know :D18:27
@HeikoShaha dudes what are you talking about? ;)18:27
@HeikoSim just doing my stuff18:27
sanujlambday: how18:27
@HeikoSits only a side job18:27
@HeikoSso all good :)18:27
@HeikoSyou gotta be good at context switching ;)18:28
@HeikoSlisitsyn: ^ you like those right?18:28
sanujhow karlnapf == clark kent18:28
@HeikoSlambday: I wanna be the evil guy18:28
@HeikoSnot the nerd ;)18:28
@HeikoSblast things18:28
sanujHeikoS: i read demis hassabis did his phd from the same place as yours18:29
sanujhe started deepmind18:29
@HeikoSsanuj: not phd, post doc18:29
@HeikoShe worked here back when I first started18:29
@lambdayHeikoS: haha that's the rebel in you.. you're more batman then18:29
@HeikoShalf of deepmind's core is from gatsby18:29
@HeikoSbut these days they are big, so more people from outside18:30
@lambdaysanuj: just kidding..18:30
@wikinglambday: heheh i would never take job18:30
@wikingnext to a phd18:30
@HeikoSthe go guy, david silver, he used to be at UCL.18:30
@HeikoSI had courses with him in my masters18:30
@wikingthat's my 2 cents18:30
@HeikoS(where he talked about go in fact)18:30
sanujHeikoS: sweet18:30
@wiking(i did that for the last 1.5 year... it's a fucking disastear18:30
@HeikoSwiking: me neither O:-)18:30
@lambdaywiking: haha you're done with yours already18:30
sanujHeikoS: i was planning to watch his reinforcement learning lectures18:30
@HeikoSthey are good18:31
@wikinglambday: indeed but that wasn't fun (the end)18:31
sanujHeikoS: i am sort of looking for things i can do research in.....mainly applications of ml18:31
@HeikoSsanuj: good to look atound for a bit18:32
@HeikoSbefore deciding what to do18:32
@HeikoSwiking: I think it depends on a lot on the context ... kind of works for me, but I would not reccomend in general18:32
@wikingheheh i would seriously first take about 6-12 months18:32
@wikingwhether you really *really* want a phd18:33
@wikingHeikoS: you work part time ?18:33
@HeikoSyeah agreed18:33
@lambdaywiking: that's what oracle offered me ... 24 months to be exact!18:33
@HeikoSwiking: yes18:33
@wikingHeikoS: related?18:33
sanujwiking: is big deal18:33
@HeikoSwiking: in a weird way18:33
@wikingHeikoS: ok so not :)18:33
@HeikoSwiking: haha, yes18:33
sanuji was am looking for long research interns for now18:33
@wikingHeikoS: for me the problem was the writing part18:34
@HeikoSwiking: what did you do?18:34
@HeikoSwiking: ah yes18:34
@HeikoSI wanna do that without job definitely18:34
@wikingi couldn't cope with writing thesis + work18:34
@wikingbecause work was always more interesting18:34
@HeikoSwiking: many people here have the same problem18:34
@HeikoSyeah I can see that18:34
@wikingthan writing up things that you've already done18:34
@HeikoSI wanna start wirting soon actually18:34
@HeikoSto avoid this problem18:34
@wikingyeah you cannot start early enough with writing18:34
@wiking :D18:34
@HeikoSwiking: but there is always things to do for shogun, so no time ;)18:35
@wikingi did that18:35
@wikingremember when i was working a lot on the whole refactor18:35
@wikingof the cmake?18:35
@HeikoSwiking: but shogun is also nice18:35
@HeikoSwiking: yes, fucking suicidal18:35
@wikingback in the summer 201318:35
@HeikoSbut good in the end18:35
@lambdayah good old days18:36
@HeikoSpeople start touching the cmake files18:36
@wikingok so that was the time i 'started' to write up18:36
@HeikoSas you predicted18:36
@lambdaywe had that configure script18:36
@wikingso the cmake refactor18:36
@wikingwas the procrastrination18:36
@wikingof my thesis writeup18:36
@wikingtrue story18:36
@lambdaywiking: haha really18:36
@HeikoSnice one, at least it was useful18:37
@wikingyeah but you dont wanna know18:37
@wikinghow long it took18:37
@wikingto write up18:37
@HeikoSI dont wanna know, you are right, haha ;)18:37
@HeikoSit would just scare me so dont tell18:37
@wikingso start writing18:37
ShoGaYOf all the friends I've had... you're the first.18:38
@lambdayShoGaY can talk!18:38
@HeikoSI gotta go now guys, see you later! :)18:38
@lambdayHeikoS: see you!18:38
sanujHeikoS: see ya18:38
@HeikoSlambday: let me know when I can start using the testing interface.,,...excited to write an example18:38
ShoGaYThe auxiliary power's out, and they spilled my cocktail.18:38
@lambdayHeikoS: absolutely! hopefully I'll merge the whole thing by today18:39
ShoGaYHe's pending for a bending...18:39
ShoGaYThis calls for a party, baby. I'm ordering 100 kegs, 100 hookers and 100 Elvis impersonators that aren't above a little hooking should the occasion arise.18:39
-!- HeikoS [] has quit [Ping timeout: 246 seconds]18:43
-!- lambday [6a33e882@gateway/web/freenode/ip.] has quit [Ping timeout: 252 seconds]18:44
-!- lambday [6a33e882@gateway/web/freenode/ip.] has joined #shogun18:46
-!- mode/#shogun [+o lambday] by ChanServ18:46
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 246 seconds]19:16
ShoGaYThose could be anyone's thoughts, fat ass!19:39
-!- livus [~suvil@] has quit [Quit: Leaving.]19:50
-!- livus [~suvil@] has joined #shogun19:51
-!- sanuj [0e8bc402@gateway/web/freenode/ip.] has quit [Ping timeout: 252 seconds]20:01
-!- shogun-notifier- [] has quit [Quit: transmission timeout]20:10
-!- sanuj [0e8bc402@gateway/web/freenode/ip.] has joined #shogun20:17
-!- sonne|osx [] has joined #shogun20:34
-!- sonne|osx [] has quit [Quit: sonne|osx]20:55
-!- sanuj [0e8bc402@gateway/web/freenode/ip.] has quit [Quit: Page closed]21:06
-!- livus [~suvil@] has quit [Quit: Leaving.]21:16
-!- shubham____ [uid57367@gateway/web/] has joined #shogun21:26
-!- shubham____ [uid57367@gateway/web/] has quit [Quit: Connection closed for inactivity]23:27
--- Log closed Tue Mar 22 00:00:40 2016