Open in new window / Try shogun cloud
--- Log opened Thu Sep 29 00:00:41 2011
-!- Guest56770 [~sonne@f053046143.adsl.alicedsl.de] has quit [Ping timeout: 252 seconds]01:23
-!- Netsplit *.net <-> *.split quits: @sonney2k, CIA-306:38
-!- Netsplit over, joins: CIA-3, @sonney2k06:44
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun10:01
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun15:32
blackburnsonne|work: around?15:33
sonne|workyes?15:58
sonne|workwassup?15:58
sonne|workblackburn: ^15:58
blackburnsonne|work: I've been thinking about integrating superLU-MT16:07
blackburnbut the problem is that debian have no superlu-mt package16:08
sonne|workwhat is MT?16:08
blackburnmultithreaded16:08
sonne|workk16:08
sonne|workwhy not16:08
blackburnsonne|work: so the question is how to integrate it16:09
blackburncan we integrate the codebase?16:09
sonne|worknot integrate16:09
sonne|workeveryone that has it installed can use it16:09
sonne|workotherwise not16:09
blackburnhmm okay16:09
blackburnsonne|work: another question16:09
blackburnI need to separate my 'toolbox' somehow16:10
blackburnare you against it in any means?16:10
sonne|workseparate what / how?16:11
blackburnsonne|work: I guess I have to make it compile-able even without shogun16:12
sonne|workthat is a lot of work...16:12
blackburnsonne|work: I don't need to separate it completely16:12
sonne|workI am certainly not a big fan of this16:12
sonne|workwhat you could do is require libshogun of some version and provide the extensions for that16:13
blackburnah yes!16:13
blackburnwhy not to require libshogun16:13
blackburnsonne|work: I'm not sure if I really need to do that, but it can't be yet another paper on shogun16:14
sonne|workthat is clear - but it is not a paper about shogun but about your dimred methods with which you extented shogun16:15
blackburnsonne|work: and the last question I've got today: would you mind me implement lapack/arpack wrappers not only for double but for float too?16:15
sonne|workfeel free16:16
blackburnsonne|work: I guess we will try to treat it as shogun extending16:16
sonne|workof course16:16
sonne|workI mean we already came to that conclusion or?16:16
blackburnsonne|work: hope that will work16:16
sonne|worklike the weka guys did with kepler16:16
sonne|worketc16:16
blackburnsonne|work: yeah okay16:17
sonne|workwe had like 3-4 jmlr submissions that extended weka somehow16:17
sonne|workor where tools that used weka internally16:17
blackburnsonne|work: today I came up with yet another algorithm extension16:18
blackburnkernel ltsa :)16:18
blackburnkind of innovative hehe16:18
blackburnsonne|work: can I implement some tests in say libshogun/lsdrt/?16:21
blackburnnot right in libshogun?16:21
blackburnbtw I think it could be good idea to develop some tests in subfolders16:22
blackburnlike /svm/ or so16:22
blackburnsonne|work: you aren't feeling happy with our effort to get published :)16:22
sonne|workwhy not examples/libshogun16:23
sonne|workno no I am perfectly fine with it16:23
sonne|workjust not too much additional effort to separate from shogun16:23
blackburnsonne|work: I need as much as needed to get published :)16:25
sonne|workreviewers will tell you16:26
blackburnokay16:29
blackburnsonne|work: can you send me some references to papers describing multiclass svms?16:34
blackburnwhen you have some time for that16:34
blackburnI ve found http://cmp.felk.cvut.cz/~xfrancv/papers/Franc-MultiSVM-ICPR02.pdf is it worth to study?16:36
sonne|workgtg16:36
sonne|workcu16:36
blackburnsee you16:37
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed]17:28
-!- mrsrikanth [~mrsrikant@59.92.74.231] has joined #shogun19:46
-!- blackburn [~blackburn@188.168.5.42] has joined #shogun20:10
-!- Guest56770 [~sonne@g229061048.adsl.alicedsl.de] has joined #shogun21:00
-!- mrsrikanth [~mrsrikant@59.92.74.231] has quit [Quit: Leaving]22:53
--- Log closed Fri Sep 30 00:00:45 2011