--- Log opened Mon Aug 06 00:00:17 2012 | ||
@sonney2k | for linear svms yes for kernel svms dot() won't change | 00:00 |
---|---|---|
@sonney2k | ohh it does - you also need to use bit shifts | 00:01 |
blackburn | but it is probably the same | 00:01 |
blackburn | it would make sense if I had no memory but it fits | 00:01 |
@sonney2k | depends - less mem req. may mean better cache usage | 00:04 |
@sonney2k | and you will have much less mem accesses too | 00:04 |
blackburn | sonney2k: unfortunately it stopped on 90% :( | 00:04 |
-!- n4nd0 [~nando@198.Red-2-137-33.dynamicIP.rima-tde.net] has joined #shogun | 00:06 | |
n4nd0 | hi blackburn | 00:06 |
blackburn | hey | 00:06 |
n4nd0 | let me ask you | 00:07 |
n4nd0 | my compile log looks pretty ugly | 00:07 |
n4nd0 | full of warnings related to Eigen | 00:07 |
n4nd0 | something local I guess | 00:07 |
n4nd0 | lot of things like this one, give me a sec | 00:07 |
blackburn | we all have it :) | 00:07 |
n4nd0 | http://pastebin.com/QjzNk4XY | 00:08 |
n4nd0 | aham | 00:08 |
n4nd0 | doesn't it look pretty bad? | 00:09 |
blackburn | no not really | 00:10 |
n4nd0 | mmm ok | 00:10 |
blackburn | I wanted to fix ithat but forgot | 00:10 |
n4nd0 | at least pretty long | 00:10 |
n4nd0 | I got a bit concerned because I run some examples but didn't get anything on screen | 00:11 |
blackburn | the key is to make it -isystem instead of -I | 00:11 |
n4nd0 | something fishy with SG_PRINT I guess | 00:11 |
n4nd0 | I thought it could be related | 00:11 |
blackburn | no should not | 00:11 |
n4nd0 | can you please tell me if your SG_PRINT works fine even with those warnings around? | 00:12 |
blackburn | yes, it works | 00:12 |
n4nd0 | ok, thank you | 00:12 |
blackburn | n4nd0: you could get warnings off for now | 00:13 |
blackburn | just open .config | 00:13 |
blackburn | find -I of eigen | 00:13 |
blackburn | and make it -isystem instead | 00:13 |
n4nd0 | ok, I will do that after my current compilation | 00:13 |
n4nd0 | thanks! | 00:13 |
blackburn | I can't stop looking at that cat | 00:21 |
blackburn | http://cs301712.userapi.com/v301712029/5979/A_hDGJZepqk.jpg | 00:21 |
@sonney2k | blackburn, it has your eyes :D | 00:23 |
n4nd0 | good night guys | 01:19 |
-!- n4nd0 [~nando@198.Red-2-137-33.dynamicIP.rima-tde.net] has quit [Quit: leaving] | 01:35 | |
-!- gsomix [~gsomix@109.169.130.0] has quit [Ping timeout: 272 seconds] | 01:43 | |
-!- blackburn [~blackburn@109.226.80.43] has quit [Quit: Leaving.] | 03:29 | |
shogun-buildbot_ | build #51 of nightly_default is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/51 | 03:37 |
-!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Quit: Page closed] | 05:14 | |
-!- gsomix [~gsomix@109.169.255.212] has joined #shogun | 07:24 | |
-!- gsomix [~gsomix@109.169.255.212] has quit [Quit: Ex-Chat] | 08:34 | |
-!- pluskid [~pluskid@108.171.196.83] has joined #shogun | 10:55 | |
-!- K0stIa [~kostia@2001:718:2:1634:4a5b:39ff:fe99:cc49] has joined #shogun | 11:05 | |
-!- K0stIa [~kostia@2001:718:2:1634:4a5b:39ff:fe99:cc49] has left #shogun [] | 11:07 | |
-!- uricamic [~uricamic@2001:718:2:1634:c92:f823:ac7e:c99f] has joined #shogun | 11:42 | |
-!- blackburn [~blackburn@109.226.80.43] has joined #shogun | 12:26 | |
-!- pluskid [~pluskid@108.171.196.83] has quit [Ping timeout: 246 seconds] | 12:27 | |
-!- pluskid [~pluskid@111.120.59.211] has joined #shogun | 12:44 | |
-!- blackburn [~blackburn@109.226.80.43] has quit [Quit: Leaving.] | 13:52 | |
-!- K0stIa [~kostia@2001:718:2:1634:4a5b:39ff:fe99:cc49] has joined #shogun | 16:13 | |
-!- K0stIa [~kostia@2001:718:2:1634:4a5b:39ff:fe99:cc49] has left #shogun [] | 16:23 | |
-!- Netsplit *.net <-> *.split quits: pluskid | 16:53 | |
-!- uricamic [~uricamic@2001:718:2:1634:c92:f823:ac7e:c99f] has quit [Quit: Leaving.] | 16:55 | |
-!- Netsplit over, joins: pluskid | 16:55 | |
-!- pluskid [~pluskid@111.120.59.211] has quit [Ping timeout: 240 seconds] | 17:08 | |
-!- pluskid [~pluskid@108.171.196.83] has joined #shogun | 17:08 | |
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has joined #shogun | 17:27 | |
-!- heiko [~heiko@dhcp-189-122.internal.eduroam.ucl.ac.uk] has joined #shogun | 17:34 | |
-!- pluskid [~pluskid@108.171.196.83] has quit [Quit: Leaving] | 17:41 | |
heiko | wiking, around? | 18:00 |
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has quit [Quit: Page closed] | 18:08 | |
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has joined #shogun | 18:53 | |
-!- KMcQuisten_ [d8338942@gateway/web/freenode/ip.216.51.137.66] has joined #shogun | 18:55 | |
KMcQuisten_ | silly webchat | 18:55 |
KMcQuisten_ | why do you torment me so? | 18:55 |
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has quit [Ping timeout: 245 seconds] | 18:57 | |
-!- blackburn [~blackburn@109.226.80.43] has joined #shogun | 19:09 | |
-!- n4nd0 [~n4nd0@198.Red-2-137-33.dynamicIP.rima-tde.net] has joined #shogun | 19:14 | |
KMcQuisten_ | Hey there blackburn | 19:15 |
blackburn | KMcQuisten_: hey I just saw your mail | 19:15 |
KMcQuisten_ | Righto | 19:16 |
blackburn | no, gsomix was a little away | 19:16 |
KMcQuisten_ | *grumble* | 19:16 |
KMcQuisten_ | Allrighty. | 19:17 |
blackburn | KMcQuisten_: actually all method calls increase reference count with director | 19:24 |
blackburn | so even some debug stuff makes it increase with no limits :D | 19:24 |
KMcQuisten_ | Zoinks | 19:24 |
KMcQuisten_ | That's a bit of a problem, isn't it? | 19:25 |
KMcQuisten_ | :) | 19:25 |
blackburn | KMcQuisten_: yes and in your example something makes features to be deleted :) | 19:25 |
KMcQuisten_ | Very odd. | 19:25 |
blackburn | kind of shootin' leg | 19:25 |
KMcQuisten_ | Yeah I hope gsomix can figure it out soon. I have a whole host of custom kernels that I want to use the directorkernel framework for. | 19:26 |
blackburn | KMcQuisten_: okay I reached him so we will try to figure it out | 19:30 |
KMcQuisten_ | Awesome. Thanks so much, guys. | 19:31 |
blackburn | KMcQuisten_: try to call __disown__() on each created director kernel | 19:44 |
blackburn | before using | 19:44 |
blackburn | I have to reconfigure shogun meanwhile | 19:44 |
KMcQuisten_ | WHen should I call it? | 19:46 |
KMcQuisten_ | Before I pass it to the machine with set_kernel()? OR after? | 19:49 |
blackburn | KMcQuisten_: after creation | 19:50 |
blackburn | when you construct combined kernel I think | 19:50 |
KMcQuisten_ | Is it a method call from the kernel object, or CombinedKernel.__disown__(directorkernel)? | 19:51 |
KMcQuisten_ | or is disown a global function | 19:51 |
blackburn | KMcQuisten_: kernel object | 19:55 |
blackburn | x.__disown__() | 19:55 |
KMcQuisten_ | Alrighty | 19:55 |
-!- heiko [~heiko@dhcp-189-122.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 19:57 | |
blackburn | uh finally compiled with directors | 19:57 |
blackburn | it goes pretty slow we need to speed up it somehow :D | 19:58 |
KMcQuisten_ | Ok, that fixed that segfault. Now, what about other situations where I'm creating, using and destroying a bunch of director kernels, but not in a combinedkernel? Do I needto call disown on any director kernel that gets passed to a regular shogun object? | 19:58 |
blackburn | yeah example works here too | 19:59 |
blackburn | KMcQuisten_: yeah each time you create anything director - disown it | 19:59 |
KMcQuisten_ | Disown it after i pass it to another object, yes? If I disown it right away won't it be deleted? | 19:59 |
blackburn | KMcQuisten_: well yeah, disown it before use in any shogun object | 20:02 |
blackburn | I am unsure if that can be done automaticall | 20:02 |
blackburn | y | 20:02 |
KMcQuisten_ | I can do an isinstance check in my code for that | 20:03 |
blackburn | I will check whether that can be done automatically | 20:03 |
KMcQuisten_ | Ok | 20:03 |
-!- heiko [~heiko@dhcp-189-122.internal.eduroam.ucl.ac.uk] has joined #shogun | 20:11 | |
-!- heiko [~heiko@dhcp-189-122.internal.eduroam.ucl.ac.uk] has quit [Client Quit] | 20:15 | |
-!- heiko [~heiko@dhcp-189-122.internal.eduroam.ucl.ac.uk] has joined #shogun | 20:15 | |
-!- n4nd0 [~n4nd0@198.Red-2-137-33.dynamicIP.rima-tde.net] has quit [Quit: Ex-Chat] | 20:20 | |
-!- KMcQuisten_ [d8338942@gateway/web/freenode/ip.216.51.137.66] has quit [Quit: Page closed] | 20:27 | |
-!- heiko [~heiko@dhcp-189-122.internal.eduroam.ucl.ac.uk] has left #shogun [] | 20:28 | |
@sonney2k | blackburn, hmmhh would be better to discus the stuff KMcQuisten runs into on the mailinglist than in privat | 21:38 |
@sonney2k | e | 21:38 |
blackburn | sonney2k: yeah | 21:43 |
blackburn | sonney2k: did you know about that issue btw? | 21:43 |
@sonney2k | blackburn, I don't even know now and no one else except you does :D | 21:50 |
@sonney2k | blackburn, do we have a function in evaluation to get the confusion matrix? | 21:50 |
blackburn | sonney2k: yes, multiclassaccuracy has one | 21:51 |
@sonney2k | blackburn, because https://github.com/shogun-toolbox/shogun/pull/679/files#L16R66 | 21:52 |
@sonney2k | I guess that one shoudl use it then? | 21:52 |
blackburn | hah | 21:52 |
blackburn | yes pluskid came with his own solution again | 21:52 |
@sonney2k | ok told him | 21:56 |
@sonney2k | blackburn, do you know if n4nd0 fixes up matrix features to be more general / whether he introduces SGMatrixList as new data type to not have yet another type*, int, int issue later? | 21:57 |
@sonney2k | blackburn, ahh and do you know where gsomix is hiding :D | 21:57 |
blackburn | no idea | 21:57 |
blackburn | about first | 21:57 |
blackburn | gsomix is available in jabber | 21:57 |
@sonney2k | it is pretty bad that we always get *huge* PRs :( | 21:58 |
@sonney2k | blackburn, I don't know about you but really carefully doing code review for these big monsters is a pain so I only do it superficially... | 21:59 |
blackburn | sonney2k: yeah it can be hard sometimes | 22:00 |
@sonney2k | pluskids PR has ~1700 lines n4nd0s 2600... | 22:05 |
@sonney2k | blackburn, ohh I see your eyes in github | 22:06 |
* sonney2k files hypnotized | 22:06 | |
blackburn | sonney2k: :D | 22:06 |
blackburn | sonney2k: isn't that cat awesome? | 22:06 |
@sonney2k | blackburn, yes | 22:08 |
@sonney2k | blackburn, did you receive the github love email | 22:08 |
blackburn | sonney2k: yeah | 22:08 |
blackburn | reading | 22:08 |
-!- n4nd0 [~nando@92.Red-2-137-9.dynamicIP.rima-tde.net] has joined #shogun | 22:22 | |
wiking | bijuuuu | 22:30 |
blackburn | wiking: what's that about?:) | 22:31 |
wiking | dunno | 22:34 |
wiking | the mars thingy has landed todaybtw :) | 22:35 |
blackburn | wiking: yeah pretty funny for me to compare that with inquisition taking place in russia | 22:36 |
-!- gsomix [~gsomix@80.234.52.32] has joined #shogun | 22:36 | |
gsomix | good evening | 22:37 |
blackburn | wiking: btw I was playing around with freaks and they are pretty interesting | 22:38 |
blackburn | wiking: I'd like to use opencv to compute features now | 22:38 |
wiking | blackburn: eheheh told you | 22:38 |
blackburn | rewriting such things is way too slow | 22:38 |
blackburn | sure | 22:38 |
wiking | opencv is good enough for that | 22:38 |
wiking | nothing else though :) | 22:38 |
blackburn | it is ugly actually but nothing better | 22:38 |
wiking | indeed | 22:39 |
blackburn | wiking: so are we going to use BoW? | 22:39 |
wiking | well since i have the code for it | 22:41 |
wiking | we could | 22:42 |
blackburn | wiking: I tried freak on road signs with one keypoint at center of image | 22:42 |
wiking | how was it? | 22:43 |
blackburn | not the best descriptor for sure but ~90% of accuracy can be reached | 22:43 |
blackburn | actually it would be better for a few keypoints I think | 22:44 |
blackburn | wiking: what I do not understand is alex's comment | 22:44 |
blackburn | I mean authors state it works.. | 22:45 |
blackburn | one reason to try freak is that it is terrible fast | 22:45 |
wiking | hehhehe | 22:48 |
wiking | well can't we do some dense freak ? | 22:48 |
wiking | or at least yeah we should do several keypoints for bow | 22:49 |
blackburn | wiking: what is dense here? | 22:49 |
wiking | well good question | 22:51 |
blackburn | wiking: you mean some grid of keypoints? | 22:52 |
blackburn | why to do that? in these pics I thought we need a few keypoints | 22:52 |
blackburn | wiking: what is your bow code based on? | 23:06 |
wiking | blackburn: that's fully java+hadoop+mahout code | 23:10 |
blackburn | wiking: how do you compute keypoints/descriptors? | 23:10 |
wiking | blackburn: that's coming from the opencv code in my case | 23:10 |
blackburn | wiking: do you think we can optimize it for binary descriptors? | 23:11 |
wiking | well | 23:12 |
wiking | afaik we could | 23:12 |
wiking | but we should just run test asap | 23:12 |
blackburn | wiking: what's actual deadline btw? | 23:12 |
-!- n4nd0 [~nando@92.Red-2-137-9.dynamicIP.rima-tde.net] has quit [Ping timeout: 252 seconds] | 23:20 | |
CIA-18 | shogun: Sergey Lisitsyn master * rfb93440 / (19 files in 2 dirs): Crossvalidation support for SLEP based multitask - https://github.com/shogun-toolbox/shogun/commit/fb93440c21ad5d7a42b6f2f178afc123166270cb | 23:34 |
shogun-buildbot_ | build #161 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/161 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 23:41 |
blackburn | oh | 23:41 |
blackburn | okay | 23:41 |
--- Log closed Tue Aug 07 00:00:17 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!