--- Log opened Wed Feb 08 00:00:15 2017 | ||
-!- lbahl [b989122b@gateway/web/freenode/ip.185.137.18.43] has joined #shogun | 00:18 | |
-!- lbahl [b989122b@gateway/web/freenode/ip.185.137.18.43] has quit [Ping timeout: 260 seconds] | 01:18 | |
-!- ShellcatZero [~ShellcatZ@cpe-76-93-136-67.san.res.rr.com] has quit [Ping timeout: 240 seconds] | 01:29 | |
-!- ShellcatZero [~ShellcatZ@cpe-76-93-136-67.san.res.rr.com] has joined #shogun | 01:30 | |
-!- suhas2go [uid201652@gateway/web/irccloud.com/x-qvmfnjcbbnefyhkr] has joined #shogun | 05:03 | |
@sukey | New Commit "[WIP] refactor cmake project | 05:53 |
---|---|---|
@sukey | add support for building modular interfaces depending on an external | 05:53 |
@sukey | prebuilt libshogun" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/33ae8cbc6cdf3f6eccdcb53aa6e9c71578f06f7e | 05:53 |
-!- travis-ci [~travis-ci@ec2-54-205-207-228.compute-1.amazonaws.com] has joined #shogun | 06:30 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/199493512 | 06:30 |
-!- travis-ci [~travis-ci@ec2-54-205-207-228.compute-1.amazonaws.com] has left #shogun [] | 06:30 | |
-!- praisethemoon [~praisethe@197.3.11.178] has joined #shogun | 09:03 | |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-ewnbugmgpmxrxiik] has joined #shogun | 09:32 | |
-!- rcurtin_ [~ryan@knife.lugatgt.org] has joined #shogun | 10:43 | |
-!- Netsplit *.net <-> *.split quits: rcurtin | 10:47 | |
@sukey | New Commit "[WIP] refactor cmake project | 11:47 |
@sukey | add support for building modular interfaces depending on an external | 11:47 |
@sukey | prebuilt libshogun" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/3d181accabed6b8a5f1fd830569d0e09668ef7fe | 11:47 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 11:47 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:47 | |
@wiking | HeikoS, fuck fuck kcuf | 11:53 |
-!- travis-ci [~travis-ci@ec2-54-211-176-33.compute-1.amazonaws.com] has joined #shogun | 12:26 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/199566472 | 12:26 |
-!- travis-ci [~travis-ci@ec2-54-211-176-33.compute-1.amazonaws.com] has left #shogun [] | 12:26 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 12:29 | |
-!- CaBa_ [~Diu7saig@lounge.imp.fu-berlin.de] has quit [Quit: Changing server] | 12:54 | |
-!- CaBa [~Diu7saig@unaffiliated/caba] has joined #shogun | 12:54 | |
-!- abhinav [~abhinav@103.211.52.16] has joined #shogun | 14:43 | |
abhinav | Ping : Wiking | 14:46 |
@wiking | pong | 14:47 |
abhinav | Hey wiking I was working on adding Scala | 14:47 |
abhinav | https://github.com/shogun-toolbox/shogun/pull/3606/files | 14:48 |
abhinav | I have made some changes for detecting scala-library.jar | 14:48 |
abhinav | Now I am using find_jar() function | 14:48 |
abhinav | Is this ok ? | 14:48 |
@wiking | does it work for you? | 14:49 |
abhinav | Yeah it does | 14:49 |
@wiking | ok col | 14:49 |
@wiking | cool | 14:49 |
@wiking | then it's good | 14:49 |
@wiking | but you are still missing | 14:49 |
@wiking | crucial part of a Find<Target>.cmake file | 14:49 |
@wiking | i told you | 14:50 |
abhinav | I have added sbt to the DockerFile | 14:50 |
@wiking | no | 14:50 |
@wiking | i'm talking about | 14:50 |
@wiking | Find<Target>.cmake | 14:50 |
abhinav | Apart from this you told me mark_as_advanced() | 14:51 |
abhinav | was old school | 14:51 |
@wiking | find_package_handle_standard_args | 14:51 |
@wiking | mark_as_advanced | 14:51 |
@wiking | these are still missing | 14:51 |
@wiking | i told you about find_package_handle_standard_args | 14:51 |
@wiking | right? | 14:51 |
@wiking | make sure that you extract in the cmake file | 14:51 |
@wiking | the version of the scala | 14:51 |
@wiking | so later we can require a given version of scala | 14:52 |
abhinav | Oh I misunderstood a little | 14:53 |
abhinav | Hey wiking. Is this fine ? | 15:54 |
abhinav | if(Scala_SCALA_EXECUTABLE) | 15:54 |
abhinav | execute_process(COMMAND ${Scala_SCALA_EXECUTABLE} -version | 15:54 |
abhinav | RESULT_VARIABLE SCALA_SEARCH_SUCCESS | 15:54 |
abhinav | ERROR_VARIABLE SCALA_VERSION | 15:54 |
abhinav | OUTPUT_STRIP_TRAILING_WHITESPACE) | 15:54 |
abhinav | if( SCALA_SEARCH_SUCCESS ) | 15:54 |
abhinav | message( FATAL_ERROR "Error executing scala -version" ) | 15:54 |
abhinav | endif() | 15:54 |
abhinav | endif() | 15:54 |
abhinav | include(FindPackageHandleStandardArgs) | 15:54 |
abhinav | if (CMAKE_VERSION LESS 2.8.3) | 15:54 |
abhinav | find_package_handle_standard_args(SCALA DEFAULT_MSG SCALA_VERSION) | 15:54 |
abhinav | else () | 15:54 |
abhinav | find_package_handle_standard_args(SCALA REQUIRED_VARS SCALA_VERSION) | 15:54 |
@wiking | abhinav, please use pastebin.com | 15:54 |
abhinav | endif () | 15:54 |
abhinav | mark_as_advanced( | 15:54 |
-!- abhinav was kicked from #shogun by wiking [abhinav] | 15:54 | |
abhinav | Scala_SCALA_EXECUTABLE | 15:54 |
abhinav | Scala_SCALAC_EXECUTABLE | 15:54 |
abhinav | Scala_JAR_EXECUTABLE | 15:54 |
abhinav | ) | 15:54 |
-!- abhinav [~abhinav@103.211.52.16] has joined #shogun | 15:55 | |
abhinav | Hey wiking : Sorry | 15:55 |
@wiking | abhinav, pastebin.com | 15:56 |
abhinav | Yeah I'll never do this again | 15:56 |
abhinav | http://pastebin.com/abku89gZ | 15:58 |
abhinav | Just wanna say that output is being stored in stderr and not in stdout | 15:59 |
@wiking | mmm | 16:03 |
@wiking | abhinav, please read manual of cmake | 16:03 |
@wiking | find_package_handle_standard_args(SCALA REQUIRED_VARS SCALA_VERSION) | 16:03 |
@wiking | is wrong | 16:03 |
@wiking | https://cmake.org/cmake/help/v3.0/module/FindPackageHandleStandardArgs.html | 16:03 |
abhinav | Yeah I'll add the variables. | 16:06 |
abhinav | And the version being accessed using ERROR_VARIABLE shouldn't be a problem | 16:07 |
abhinav | ? | 16:07 |
@wiking | is that working/ | 16:10 |
@wiking | i mean sorry i'm not a compiler | 16:10 |
@wiking | use your machine | 16:10 |
@wiking | to do these tests | 16:10 |
abhinav | Yeah it's working and I am getting the output | 16:10 |
@wiking | no it's not that variables that are missing | 16:10 |
@wiking | in | 16:10 |
@wiking | find_package_handle_standard_args | 16:10 |
@wiking | you are using it wrong | 16:10 |
@wiking | if you set version | 16:11 |
@wiking | then you should do ... VERSION_VAR SCALA_VERSION | 16:11 |
@wiking | etc | 16:11 |
@wiking | plz read man | 16:11 |
abhinav | Sorry. I'll just first go through and make sure everything is right | 16:11 |
abhinav | Thanks | 16:11 |
-!- rcurtin_ is now known as rcurtin | 16:14 | |
-!- abhinav [~abhinav@103.211.52.16] has quit [Quit: Leaving] | 16:26 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 16:58 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:58 | |
@HeikoS | wiking: ! | 17:05 |
@wiking | ! | 17:05 |
@HeikoS | wiking: jojo | 17:08 |
@HeikoS | wiking: I see things happening | 17:08 |
@HeikoS | wiking: btw you know about our mailing list archives? | 17:08 |
@wiking | gmane | 17:08 |
@wiking | yes | 17:08 |
@sukey | Wiki page: Home edited on shogun-toolbox/shogun by karlnapf | 17:08 |
@wiking | apparently itll come back | 17:09 |
@wiking | but dunno | 17:09 |
@wiking | maybe never | 17:09 |
@HeikoS | I see | 17:09 |
@HeikoS | link on website is dead | 17:09 |
@HeikoS | I already got like 10 emails | 17:09 |
@HeikoS | about that | 17:09 |
@HeikoS | since gsoc students write to me | 17:09 |
@HeikoS | without reading ANYTHING that I wrote in the wiki before writing | 17:10 |
@sukey | Wiki page: Google-Summer-of-Code-2017-Projects edited on shogun-toolbox/shogun by karlnapf | 17:10 |
@wiking | but they can sign up for the mailing list | 17:11 |
@wiking | an write | 17:11 |
@wiking | just the archive is dead | 17:11 |
@HeikoS | I know | 17:11 |
@HeikoS | wiking: so in terms of projects | 17:11 |
@HeikoS | wiking: you have a minute to discuss? | 17:11 |
@wiking | yeah | 17:11 |
@sukey | Wiki page: Google-Summer-of-Code-2017-Projects edited on shogun-toolbox/shogun by karlnapf | 17:11 |
@wiking | although it's 12pm | 17:11 |
@HeikoS | wiking: cool | 17:11 |
@wiking | so i'm getting tired | 17:11 |
@HeikoS | Ill be quick | 17:11 |
@wiking | but shoot | 17:11 |
@HeikoS | just give me some y/n | 17:11 |
@wiking | it's not up to your quickness | 17:11 |
@wiking | :))) | 17:11 |
@HeikoS | installation project? | 17:11 |
@wiking | it's my mind | 17:11 |
@wiking | ah yeah that is still open | 17:12 |
@wiking | although | 17:12 |
@wiking | honestly | 17:12 |
@wiking | by the time gsoc starts | 17:12 |
@wiking | it'll be done | 17:12 |
@wiking | ecause we need it | 17:12 |
@HeikoS | what about combining with other things around that | 17:12 |
@HeikoS | like maven | 17:12 |
@HeikoS | etc | 17:12 |
@HeikoS | windows installer | 17:12 |
@HeikoS | still work | 17:12 |
@wiking | yeah thats' what i meant | 17:12 |
@wiking | that'll be done by then | 17:13 |
@wiking | march 15 | 17:13 |
@HeikoS | mmh ok | 17:13 |
@wiking | maven + windows + python | 17:13 |
@HeikoS | so you think not do the project? | 17:13 |
@HeikoS | kk | 17:13 |
@HeikoS | wiking: next is unified ML interface & tags | 17:13 |
@HeikoS | I kind of want to integrate tags before gsoc | 17:13 |
@HeikoS | but this still gives room to design the core interface API | 17:14 |
@wiking | eah i mean itd be good to integrate gsoc16 stuff before gsoc17 | 17:14 |
@wiking | ;D | 17:14 |
@HeikoS | is that worth a project? | 17:14 |
@HeikoS | wiking: yeah I know ;) | 17:14 |
@wiking | mm | 17:14 |
@wiking | full proj | 17:14 |
@wiking | no i dont think so | 17:14 |
@HeikoS | the API thing might be cool to do | 17:14 |
@HeikoS | like review other packages for API ideas | 17:14 |
@HeikoS | etc | 17:14 |
@HeikoS | and then the refactor | 17:14 |
@HeikoS | and plugin stuff of course | 17:15 |
@HeikoS | wiking: so you think that should also be cancelled? | 17:15 |
@wiking | y | 17:15 |
@HeikoS | kkj | 17:15 |
@HeikoS | wiking: next one: detox 2 | 17:16 |
@HeikoS | which will be be: file writers, cereal integration, and more things (which ones) | 17:16 |
@wiking | hehe there's a lot for that :) | 17:16 |
@HeikoS | ok | 17:16 |
@wiking | cereal needs merging | 17:16 |
@HeikoS | what topics you think can go in? | 17:16 |
@wiking | so forget that | 17:16 |
@wiking | file writers? | 17:16 |
@HeikoS | wiking: and readers | 17:16 |
@wiking | what is that? | 17:16 |
@HeikoS | wiking: I had the csvreader bug in mind for example | 17:17 |
@wiking | other than that | 17:17 |
@wiking | ? | 17:17 |
@HeikoS | wiking: I guess openmp / pthread madness could go in | 17:17 |
@HeikoS | smart pointers | 17:17 |
@HeikoS | I mean generally cleaning up those IO classes | 17:18 |
@wiking | file writers readers = I had the csvreader bug in mind for example => other than that in that area? | 17:18 |
@HeikoS | the streaming stuff | 17:18 |
@HeikoS | as in streaming framework is quite a pain | 17:18 |
@wiking | yeah i mean for detox 2 there's shitload | 17:18 |
@wiking | only thing is that we need somebody who works fast | 17:18 |
@HeikoS | ok so Ill update the page and then we can go from there | 17:18 |
@HeikoS | wiking: sure, but this is now for application | 17:19 |
@HeikoS | since deadline is tomorrow and then they will start checking | 17:19 |
@wiking | ah the ideas | 17:19 |
@wiking | mmm when's the deadline? | 17:19 |
@wiking | exactemente | 17:19 |
@HeikoS | tomorrow deadline | 17:19 |
@HeikoS | wiking: ok then, lets keep detox and I will work on that | 17:19 |
@wiking | what i mean | 17:19 |
@wiking | EXACT UTC TIME | 17:19 |
@HeikoS | wiking: next one: efficiency | 17:19 |
@HeikoS | wiking: doesnt matter, we have applied already | 17:20 |
@HeikoS | this is more for when they check the ideas pages etc | 17:20 |
@wiking | February 9 16:00 UTC | 17:20 |
@wiking | yeah in know | 17:20 |
@wiking | i wanna add there stuff | 17:20 |
@wiking | before | 17:20 |
@wiking | that deadline | 17:20 |
@HeikoS | unless they freeze them at that time | 17:20 |
@wiking | but ok in the morning i can | 17:20 |
@HeikoS | I will work now | 17:20 |
@HeikoS | so efficiency, usualy suspects there is tons of stuff | 17:20 |
@HeikoS | so that stays | 17:20 |
@HeikoS | then the kaggle pipeline thing | 17:20 |
@wiking | mmm | 17:21 |
@HeikoS | that has to be written, will do today, you got comments on that one? | 17:21 |
@wiking | that could be distingueshed | 17:21 |
@HeikoS | like "make it easy to use shogun for supervised prediction" | 17:21 |
@wiking | if i could spell that'd be great | 17:21 |
@wiking | mmm so i mean applying to kaggle competitions | 17:21 |
@wiking | great | 17:21 |
@wiking | but we need a way to easily define pipelines | 17:21 |
@wiking | anyhow | 17:21 |
@HeikoS | mostly: feature extraction / stacking, model-selection, ensemble | 17:22 |
@HeikoS | parallelisation | 17:22 |
@wiking | yeah | 17:22 |
@HeikoS | but yeah can go wild with this | 17:22 |
@HeikoS | but I want it more to be filling the gaps in the framework | 17:22 |
@HeikoS | cool | 17:22 |
@HeikoS | then interfaces | 17:22 |
@HeikoS | SWIG: matlab, julia | 17:22 |
@HeikoS | js | 17:22 |
@wiking | julia = forget no swig | 17:22 |
@wiking | matlab yes | 17:22 |
@wiking | js yes | 17:23 |
@HeikoS | so matlab and js | 17:23 |
@HeikoS | and then cleaning up maybe | 17:23 |
@HeikoS | like figure out the lua problems with overloadin | 17:23 |
@wiking | yeah | 17:23 |
@HeikoS | ok | 17:23 |
@HeikoS | and finally: application | 17:23 |
@wiking | there's shitload of things to do | 17:23 |
@wiking | with interfaces | 17:23 |
@HeikoS | I have the idea to write a general proposal, that describes using shogun to solve some problem /reproduce something | 17:23 |
@wiking | ok | 17:24 |
@HeikoS | and then outcome is self contained docker | 17:24 |
@wiking | so have you read my email | 17:24 |
@wiking | about the eco people? | 17:24 |
@HeikoS | let me do | 17:24 |
@HeikoS | ah yeah | 17:24 |
@HeikoS | for example | 17:24 |
@HeikoS | thing is | 17:24 |
@HeikoS | lea has some cool ideas as well, she wanted to propose a concrete application project | 17:24 |
@HeikoS | around data-based journalism | 17:24 |
@HeikoS | i havent read that yet, but might be cool | 17:24 |
@HeikoS | it is kind of borderline in terms of relevance for us, but I think could be cool and is sort of minimal effort | 17:25 |
@HeikoS | as we can just define deliverables | 17:25 |
@HeikoS | and then its more independent than the other projects | 17:25 |
@wiking | ye | 17:25 |
@HeikoS | kk | 17:25 |
@wiking | ah yeah now i remember | 17:25 |
@wiking | we alked about this | 17:25 |
@HeikoS | yes | 17:25 |
@wiking | completely selfcontained | 17:26 |
@wiking | stuff | 17:26 |
@wiking | 'go wild' | 17:26 |
@wiking | we could have like 3 of these :D | 17:26 |
@HeikoS | I know | 17:26 |
@HeikoS | but need project ideas | 17:26 |
@wiking | kk | 17:26 |
@HeikoS | so I will write one generic | 17:26 |
@HeikoS | and one concrete | 17:26 |
@wiking | yeah i have that one | 17:26 |
@HeikoS | then we can see | 17:26 |
@wiking | and i'll try to come up with others | 17:26 |
@wiking | like | 17:26 |
@wiking | other one is this | 17:26 |
@wiking | integrate shogun with quantlib and http://opensourcerisk.org/ | 17:26 |
@wiking | aaah fuck | 17:26 |
@wiking | there it goas | 17:27 |
@wiking | goes | 17:27 |
@wiking | TYPE | 17:27 |
@wiking | :) | 17:27 |
@wiking | so type definitions | 17:27 |
@wiking | like in stan | 17:27 |
@wiking | that'd be great | 17:27 |
@HeikoS | I see | 17:27 |
@wiking | easy to define the problem | 17:27 |
@HeikoS | is that detox? | 17:27 |
@HeikoS | or on its own | 17:27 |
@wiking | involves a lot of manual labor | 17:27 |
@wiking | after the header is done | 17:27 |
@HeikoS | open risk is a cool idea | 17:27 |
@HeikoS | then in terms of algos | 17:28 |
@wiking | it's all c++ | 17:28 |
@HeikoS | I thought: time-series (Gaussian state space models), maybe approximate kernel method | 17:28 |
@HeikoS | maybe large-scale GPs | 17:28 |
@HeikoS | all depending on applicants | 17:28 |
@HeikoS | but these are the ones that we can mentor | 17:28 |
@HeikoS | and your solver thing | 17:28 |
@wiking | yep | 17:28 |
@HeikoS | KKT | 17:28 |
@wiking | oh that'd be great | 17:29 |
@wiking | :> | 17:29 |
@HeikoS | cool | 17:29 |
@HeikoS | thanks then | 17:29 |
@HeikoS | I will polish the wiki now | 17:29 |
@HeikoS | I already can tell you what will get most applicants: | 17:29 |
@HeikoS | usual suspects2 and the kaggle thing | 17:29 |
@HeikoS | wonder whether we should have more stuff on deep learning | 17:29 |
@HeikoS | but that can also go under efficiency | 17:29 |
@HeikoS | maybe autoencoder ... | 17:30 |
@HeikoS | theo is now at geometric intelligence / uber so he cannot mentor I guess | 17:30 |
@wiking | autodif | 17:30 |
@wiking | :D | 17:30 |
@HeikoS | ah yeah | 17:30 |
@HeikoS | that would be cool | 17:30 |
@wiking | but for that | 17:30 |
@wiking | we shoudl have types | 17:30 |
@wiking | :> | 17:30 |
@HeikoS | Rahul had some ideas and prototypes | 17:30 |
@HeikoS | ok Ill write that down as well | 17:30 |
@HeikoS | would be good for later | 17:30 |
@wiking | yeah | 17:31 |
@wiking | anyhow | 17:31 |
@HeikoS | cool | 17:31 |
@wiking | many many things | 17:31 |
@wiking | and fuck colpack | 17:31 |
@wiking | :D | 17:31 |
@HeikoS | yep indeed | 17:31 |
@HeikoS | that will be banned to plugin soon :) | 17:31 |
@wiking | have u seen he tf fold? | 17:32 |
@HeikoS | seen what? | 17:32 |
@wiking | https://github.com/tensorflow/fold | 17:32 |
@HeikoS | nope | 17:32 |
@HeikoS | ah nice | 17:32 |
@HeikoS | so finally somebody is doing that stuff :) | 17:33 |
@wiking | mmm | 17:34 |
@wiking | do you use | 17:34 |
@wiking | jinja2 in tests/meta | 17:34 |
@wiking | no right? | 17:34 |
@HeikoS | ehm | 17:36 |
@HeikoS | no dont think | 17:36 |
@HeikoS | but I might have forgotten | 17:36 |
@HeikoS | implicitly maybe | 17:36 |
@HeikoS | there is requirements.txt | 17:36 |
@wiking | i'm currently fully rewamping cmake | 17:39 |
@HeikoS | i see it | 17:40 |
@HeikoS | scary | 17:40 |
@wiking | well i need this | 17:40 |
@wiking | to be able to build modular interfaces | 17:40 |
@wiking | to an already installed libshogun | 17:40 |
@wiking | consequently | 17:41 |
@wiking | the ./CMakeLists.txt is only 845 lines | 17:42 |
@wiking | instead of 1241 | 17:42 |
@wiking | 68% \o/ | 17:43 |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-ewnbugmgpmxrxiik] has quit [Quit: Connection closed for inactivity] | 17:59 | |
CaBa | hey people | 18:01 |
@wiking | y0 | 18:04 |
@HeikoS | CaBa: jojo | 18:13 |
@HeikoS | wiking: thats cool ! | 18:13 |
@HeikoS | wiking: does it work already? | 18:13 |
@HeikoS | I have the feeling that this is hard ;) | 18:13 |
-!- praisethemoon [~praisethe@197.3.11.178] has quit [Ping timeout: 240 seconds] | 18:14 | |
@sukey | Wiki page: GSoC_2017_project_kaggle created on shogun-toolbox/shogun by karlnapf | 18:39 |
@sukey | Wiki page: GSoC_2017_project_kaggle edited on shogun-toolbox/shogun by karlnapf | 18:44 |
@HeikoS | wiking: you still there? | 18:44 |
@HeikoS | wiking: https://github.com/shogun-toolbox/shogun/wiki/GSoC_2017_project_kaggle | 18:44 |
@HeikoS | lisitsyn: ^ | 18:44 |
@sukey | Wiki page: Google-Summer-of-Code-2017-Projects edited on shogun-toolbox/shogun by karlnapf | 18:47 |
CaBa | is there an explanation on what "locking" means in the shogun work somewhere out there? in the wiki or so? | 18:52 |
CaBa | *world | 18:52 |
@HeikoS | CaBa: nope :) | 19:00 |
@HeikoS | but youre lucky | 19:00 |
@HeikoS | I wrote that :) | 19:01 |
CaBa | ehm - okay ^^ | 19:01 |
@HeikoS | so it is just precomputing the kernel in cross-validation | 19:01 |
CaBa | and then subsetting on the precomputed kernel matrix? | 19:01 |
@HeikoS | yep | 19:01 |
@HeikoS | it is some way of hiding all that to the outside | 19:02 |
CaBa | some kernels don't support locking iirc. what are the prerequisites? | 19:02 |
@HeikoS | http://www.shogun-toolbox.org/api/latest/classshogun_1_1CKernelMachine.html | 19:02 |
@HeikoS | this is almost the only class which implements it properly | 19:02 |
@HeikoS | locking is more a thing of CMachine | 19:03 |
CaBa | hm. i remember something in the context of string features i think where i ran into a message telling me that locking isn't supported.. | 19:04 |
@HeikoS | yeah a machine might not support it | 19:04 |
@HeikoS | mmh | 19:04 |
@HeikoS | I mean check the code | 19:04 |
CaBa | I'm using only CLibSVM | 19:04 |
CaBa | and CMKLClassification | 19:04 |
CaBa | they are derived from CKernelMachine so they should be fine i guess ;) | 19:04 |
@HeikoS | http://www.shogun-toolbox.org/api/latest/KernelMachine_8cpp_source.html | 19:05 |
@HeikoS | line 623 | 19:05 |
@HeikoS | when you lock, you can only call train_locked apply_locked after | 19:05 |
@HeikoS | not the normal ones | 19:05 |
CaBa | aah see - combined kernel | 19:05 |
@HeikoS | I think yes | 19:05 |
@HeikoS | ah | 19:05 |
@HeikoS | yeah | 19:05 |
CaBa | so CMKLClassification won't do ;) | 19:05 |
@HeikoS | nope | 19:05 |
@HeikoS | so you cannot make it precompute the kernel | 19:06 |
@HeikoS | but you can still precompute it yourself | 19:06 |
@HeikoS | (I think=) | 19:06 |
@HeikoS | CaBa: ah man | 19:06 |
@HeikoS | this looks llike we can spend a while GSoC project just on the stuff you are trying to do :) | 19:06 |
CaBa | lol | 19:06 |
CaBa | i mean technically... precomputing all the subkernels and subsetting within them should be doable, right? | 19:07 |
@HeikoS | I think | 19:07 |
CaBa | if i understand you correctly, currently the precomputed kernel matrix resides in CMachine? | 19:08 |
CaBa | CKernelMachine | 19:08 |
@HeikoS | CMachine | 19:09 |
@HeikoS | and kernel machine specialises it so yes | 19:09 |
CaBa | HeikoS: the locking just replaces the actual kernel with a custom kernel internally? | 19:11 |
@HeikoS | yep | 19:12 |
@HeikoS | and thats why we need to avoid calling the other methods (that dont know about that=) | 19:12 |
@HeikoS | so quite dirty | 19:12 |
@HeikoS | ;) | 19:12 |
@sukey | Issue #2113 "CSGObject: all parameters needed in generic base class?" closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/2113 | 19:13 |
CaBa | HeikoS: so maybe a CCombinedCustomKernel class would help bring that to CMKL and downstream? | 19:14 |
CaBa | designwise i mean... | 19:15 |
@sukey | Wiki page: GSoC_2017_detox edited on shogun-toolbox/shogun by karlnapf | 19:20 |
@sukey | Issue #1031 "Serialization/Migration framework problem" closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/1031 | 19:30 |
@sukey | Issue #869 "Register parameters for serialisation/model-selection (easy)" closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/869 | 19:30 |
@sukey | Wiki page: GSoC_2017_detox edited on shogun-toolbox/shogun by karlnapf | 19:31 |
@sukey | Wiki page: GSoC_2017_detox edited on shogun-toolbox/shogun by karlnapf | 19:33 |
@sukey | Wiki page: GSoC_2017_detox edited on shogun-toolbox/shogun by karlnapf | 19:34 |
-!- suhas2go [uid201652@gateway/web/irccloud.com/x-qvmfnjcbbnefyhkr] has quit [Remote host closed the connection] | 19:37 | |
@sukey | New Commit "[WIP] refactor cmake project | 19:38 |
@sukey | add support for building modular interfaces depending on an external | 19:38 |
@sukey | prebuilt libshogun" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/66ee2f0f43228e9bffa348b28527666db5b82fa3 | 19:38 |
lisitsyn | hey HeikoS | 19:51 |
@HeikoS | lisitsyn: jo | 19:51 |
lisitsyn | I wasn't following gsoc stuff | 19:51 |
lisitsyn | does it make sense to proof read | 19:51 |
lisitsyn | ? | 19:51 |
@sukey | Wiki page: GSoC_2017_project_fundamental_usual_suspects edited on shogun-toolbox/shogun by karlnapf | 19:51 |
@HeikoS | yes read the kaggle thing | 19:51 |
@HeikoS | I am currently working on the other descriptions | 19:52 |
@HeikoS | want to get them down before tomorrow deadline | 19:52 |
@HeikoS | as I am afraid that they clone the html then | 19:52 |
@HeikoS | so it is fair | 19:52 |
@HeikoS | lisitsyn: btw | 19:53 |
@HeikoS | we need to merge tags before gsoc | 19:53 |
@HeikoS | really need to | 19:53 |
@HeikoS | and do something about the plugins | 19:53 |
lisitsyn | ok I should really schedule some work on this | 19:54 |
@sukey | Wiki page: GSoC_2017_project_fundamental_usual_suspects edited on shogun-toolbox/shogun by karlnapf | 19:56 |
@sukey | Wiki page: Google-Summer-of-Code-2017-Projects edited on shogun-toolbox/shogun by karlnapf | 19:57 |
@sukey | Wiki page: GSoC_2017_project_swig edited on shogun-toolbox/shogun by karlnapf | 20:00 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 276 seconds] | 20:05 | |
-!- travis-ci [~travis-ci@ec2-54-211-176-33.compute-1.amazonaws.com] has joined #shogun | 20:20 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/199710727 | 20:20 |
-!- travis-ci [~travis-ci@ec2-54-211-176-33.compute-1.amazonaws.com] has left #shogun [] | 20:20 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 20:24 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Remote host closed the connection] | 21:19 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 21:35 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Ping timeout: 240 seconds] | 21:39 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 21:56 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Remote host closed the connection] | 21:58 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 21:58 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Remote host closed the connection] | 21:59 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 22:18 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 23:14 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 23:14 | |
@sukey | Wiki page: GSoC_2016_project_fundamental_usual_suspects edited on shogun-toolbox/shogun by karlnapf | 23:15 |
@sukey | Wiki page: GSoC_2016_project_fundamental_usual_suspects edited on shogun-toolbox/shogun by karlnapf | 23:17 |
@sukey | Wiki page: Google-Summer-of-Code-2017-Projects edited on shogun-toolbox/shogun by karlnapf | 23:19 |
@sukey | Wiki page: GSoC_2017_project_kaggle edited on shogun-toolbox/shogun by lgoetz | 23:27 |
@sukey | Wiki page: GSoC_2017_project_swig edited on shogun-toolbox/shogun by karlnapf | 23:30 |
@sukey | Wiki page: GSoC_2017_project_swig edited on shogun-toolbox/shogun by karlnapf | 23:32 |
@sukey | Issue #2584 "SWIG: Write simple unit tests for type maps" closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/2584 | 23:32 |
@sukey | Wiki page: GSoC_2017_project_swig edited on shogun-toolbox/shogun by karlnapf | 23:33 |
@sukey | Wiki page: GSoC_2017_project_kaggle edited on shogun-toolbox/shogun by lgoetz | 23:35 |
@sukey | Wiki page: GSoC_2016_project_approx_kernels edited on shogun-toolbox/shogun by karlnapf | 23:39 |
@sukey | Wiki page: GSoC_2015_project_deep_learning edited on shogun-toolbox/shogun by karlnapf | 23:43 |
@sukey | Wiki page: GSoC_2017_applications created on shogun-toolbox/shogun by karlnapf | 23:49 |
@sukey | Wiki page: GSoC_2017_applications edited on shogun-toolbox/shogun by karlnapf | 23:51 |
--- Log closed Thu Feb 09 00:00:17 2017 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!