--- Log opened Wed Mar 05 00:00:25 2014 | ||
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has joined #shogun | 00:02 | |
-!- HeikoS [~heiko@nat-174-158.internal.eduroam.ucl.ac.uk] has joined #shogun | 00:15 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 00:15 | |
@HeikoS | lambday: hey! | 00:17 |
---|---|---|
@HeikoS | how are things? | 00:17 |
lambday | HeikoS: hi! | 00:18 |
lambday | lol almost slept! | 00:18 |
lambday | checking out code quadratic time mmd and linear mmd | 00:18 |
@HeikoS | lambday: haha | 00:18 |
@HeikoS | lambday: probably full of bugs ;) | 00:18 |
lambday | logical bugs? | 00:19 |
@HeikoS | lambday: hopefully not, | 00:19 |
@HeikoS | lambday: but I unit tested against arthurs implementation | 00:19 |
@HeikoS | so should be ok | 00:19 |
lambday | :D | 00:19 |
lambday | btw the only unit test we have on this is for quadratic mmd | 00:20 |
@HeikoS | no linear time test? | 00:20 |
lambday | no - check tests/unit/statistic | 00:20 |
@HeikoS | lambday: there might be examples where I added assertions, from the days before unit tests | 00:20 |
lambday | HeikoS: yeah - you used those in libshogun tests thogh | 00:20 |
@HeikoS | lambday: ther are linear time mmd tests | 00:21 |
@HeikoS | TEST(LinearTimeMMD,test_linear_mmd_fixed) | 00:21 |
@HeikoS | lambday: no hsic | 00:22 |
@HeikoS | lambday: let me write an entrance task ;) | 00:22 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 1768a47 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/1768a47a5781c3bad6c1a159503a49bf22f755bd | 00:23 |
shogun-notifier- | shogun: HAVE_EIGEN3 guard changes to counter make error w/o eigen3 | 00:23 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 936868b / / (6 files): https://github.com/shogun-toolbox/shogun/commit/936868b2472d53d5fe6de0d11b1a6d7db78288e7 | 00:23 |
shogun-notifier- | shogun: removed lapack guard from LMNN files | 00:23 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 1332111 / NEWS: https://github.com/shogun-toolbox/shogun/commit/133211135212d1cbd2056ca61e73939bbb9388ec | 00:23 |
shogun-notifier- | shogun: very minor changes in NEWS | 00:23 |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 2c1f982 / src/shogun/io/MLDataHDF5File.cpp: https://github.com/shogun-toolbox/shogun/commit/2c1f982a3b9a3ae534484b70ad6bcc26c5ba74af | 00:23 |
shogun-notifier- | shogun: changed ERROR to WARNING to tackle travis/buildbot failure when mldata is in flaky state | 00:23 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 6b67046 / / (13 files): https://github.com/shogun-toolbox/shogun/commit/6b67046d95a09413937f48a1d1d05bb4eb573e76 | 00:23 |
shogun-notifier- | shogun: Merge pull request #1927 from mazumdarparijat/pca | 00:23 |
shogun-notifier- | shogun: | 00:23 |
shogun-notifier- | shogun: HAVE_EIGEN3 guard changes to counter make error w/o eigen3 | 00:23 |
lambday | argh - *rubs eyes* | 00:23 |
@HeikoS | lambday: dont forget to sleep in between | 00:24 |
lambday | HeikoS: hsic one should be same as the example where you asserted mean/vars right? | 00:24 |
lambday | hehe! | 00:24 |
@HeikoS | lambday: yeah the examples in fact should just demonstrate API and not do complicated stuff | 00:25 |
@HeikoS | lambday: all non-trivial stuff should be in the notebook, examples just for API, testing only in unit tests | 00:25 |
@HeikoS | lambday: this is all old stuff before we did things properly | 00:25 |
lambday | HeikoS: yeah - so I'll pull those assertions inside unit test then | 00:26 |
lambday | hmm | 00:26 |
lambday | HeikoS: wow my work is forking up - see https://github.com/shogun-toolbox/shogun/issues/1930 | 00:27 |
lambday | :D | 00:27 |
@HeikoS | lambday: yeah, good isnt it? | 00:27 |
lambday | yep :D | 00:27 |
@HeikoS | lambday: you could go for it if interested | 00:28 |
@HeikoS | lambday: no need to solve the whole gsoc project already ;) | 00:28 |
lambday | haha! :D | 00:28 |
lambday | lol you just mentioned what I was about to say regarding min/max | 00:35 |
@HeikoS | lambday: feel free to elaborate on that | 00:44 |
@HeikoS | you wrote all the stuff an know best | 00:44 |
@HeikoS | lambday: you could maybe guide someone in doing this | 00:45 |
@HeikoS | lambday: feel free to point someone to do this | 00:45 |
@HeikoS | lambday: since its very useful for us | 00:45 |
sunil1337 | @HeikoS: please take a look at https://github.com/shogun-toolbox/shogun/issues/1887 and tell me what exactly is needed i've create the new method | 00:47 |
lambday | HeikoS: yeah I hope this stuff would be useful - will share ideas on that whatever I can | 00:47 |
@HeikoS | lambday: ah that one.... | 00:47 |
lambday | HeikoS: btw quadratic mmd sample_null_spectrum should get rid of lapack and depend on eigen, right? changing it then | 00:49 |
@HeikoS | lambday: I guess we want a copy method and a return reference method | 00:49 |
@HeikoS | lambday: and the reference method is not possible when a subset is set | 00:49 |
lambday | HeikoS: ummm you meant sunil1337 :P | 00:49 |
@HeikoS | lambday: the problem here is if we decide to change the feature matrix | 00:49 |
@HeikoS | lambday: but maybe we should leave that for now | 00:49 |
sunil1337 | @HeikoS: yup | 00:49 |
@HeikoS | lambday: yeah totally, in fact CStatistics should get rif of the covariance method, this should just be done with eigen | 00:50 |
@HeikoS | lambday: I closed the 1887 | 00:51 |
@HeikoS | not really clear what I want there, and its not important for now | 00:51 |
-!- HeikoS [~heiko@nat-174-158.internal.eduroam.ucl.ac.uk] has left #shogun [] | 00:51 | |
sunil1337 | @HeikoS: okay! | 00:51 |
sunil1337 | @Heikos: will try something else | 00:52 |
-!- HeikoS [~heiko@nat-174-158.internal.eduroam.ucl.ac.uk] has joined #shogun | 00:52 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 00:52 | |
lambday | HeikoS: wb | 00:52 |
lambday | HeikoS: it was sunil1337 who wanted to do 1887 :P | 00:53 |
sunil1337 | @HeikoS: lol...:D | 00:53 |
lambday | sunil1337: feel free to give 1930 a shot :) | 00:53 |
@HeikoS | sunil1337: ah sorry | 00:53 |
sunil1337 | lambday: sure buddy :) | 00:53 |
@HeikoS | sunil1337: that was a stupid issue, I should not have written that ;) | 00:54 |
sunil1337 | @HeikoS: happens all the time :) good that i din't comment :p | 00:54 |
@HeikoS | sunil1337: always let us know once you are on something | 00:55 |
@HeikoS | sunil1337: 1930 is cool, but a bit larger | 00:55 |
@HeikoS | but very useful for us! | 00:55 |
@HeikoS | lambday: can guide it :) | 00:55 |
aroma | Hi. I am working on 1913. | 00:55 |
sunil1337 | @HeikoS: will see :) | 00:55 |
aroma | I was just wondering if implementing serialization would be better or isolated load/save functions | 00:56 |
@HeikoS | aroma: pls write a little comment in the issue to avoid duplicate work | 00:56 |
@HeikoS | aroma: what do you mean? | 00:56 |
@HeikoS | aroma: could you elaborate? | 00:56 |
@HeikoS | aroma: I added a comment, lets see what shell says | 00:57 |
sunil1337 | @HeikoS: will be back have to google half of the stuff :P | 00:57 |
aroma | @HeikoS: I mean if I implement serialization, we would have to stick to certain format | 00:57 |
aroma | @HeikoS: certain format for saving/storing factor graphs | 00:58 |
@HeikoS | sunil1337: eig and eigs in matlab | 00:58 |
@HeikoS | aroma: ask shell about that, kin the issue | 00:59 |
@HeikoS | in | 00:59 |
lambday | just out of curiosity, factors graphs aren't bipartite graphs, are they? | 00:59 |
@HeikoS | lambday: no | 00:59 |
aroma | @HeikoS: ok | 00:59 |
@HeikoS | just a graph where nodes are connected via factors which are arbritary functions connecting the nodes | 01:00 |
@HeikoS | lambday: joint distribution is product of all factors, normaliser is the integral of this over all nodes | 01:00 |
lambday | HeikoS: yeah I remember now... | 01:00 |
lambday | sum product algorithm | 01:01 |
@HeikoS | lambday: no | 01:01 |
@HeikoS | lambday: thats for chain graphs :) | 01:01 |
lambday | HeikoS: umm... but I remember reading about sum product algo for factor graph somewhere - cannot recollect clearly | 01:02 |
sunil1337 | lambday: refactor in what ways.. | 01:02 |
@HeikoS | lambday: usually one has to to loopy BP | 01:02 |
sunil1337 | lambday: put in a GPU based solver? | 01:03 |
@HeikoS | sunil1337: the refactoring is just about interfaces | 01:03 |
@HeikoS | sunil1337: so no GPU solver for now | 01:03 |
@HeikoS | just eigen3 | 01:03 |
@HeikoS | but this is about unifying | 01:03 |
lambday | sunil1337: yeah - if you have a look its currently only made for computing min/max eigenvalues - but we may need to change that | 01:04 |
@HeikoS | lambday, sunil1337 the current class should be SparseEigenSolver or so | 01:05 |
@HeikoS | lambday: and it should be possible to specify the n biuggest/smallest eigenvalues rather than one | 01:05 |
lambday | HeikoS: well, EigenSolver has one implementation for dense as well | 01:05 |
@HeikoS | lambday: and the dense eigensolver should just compute all | 01:05 |
@HeikoS | lambday: ah I see | 01:05 |
@HeikoS | lambday: but the dense one needs an interface to compute all eigenvalues | 01:05 |
lambday | HeikoS: yeah! | 01:05 |
sunil1337 | lambday: okay.. | 01:06 |
lambday | HeikoS: I don't think we'll be needing all eigenvalues for sparse as well, or just have the interface there, keeping it virtual may be :-/ | 01:06 |
@HeikoS | no these should be two abstract classes | 01:06 |
lambday | HeikoS: I mean, for the purpose that wiking mentioned | 01:06 |
@HeikoS | one for all eigenvalues and one for the n largest/smallest | 01:06 |
@HeikoS | lambday: and all eigensolver should have convinience constructors that allow to give them SGMatrix SGSparseMatrix directly, could you make that happen (dont do it yourself, just guide sunil1337 :) | 01:07 |
lambday | HeikoS: oh currently it all works on linear operators, doesn't it | 01:08 |
lambday | yeah | 01:08 |
@HeikoS | lambday: exactly, but its a bit annyoing sometimes to create all those classes | 01:08 |
lambday | true that! | 01:08 |
@HeikoS | lambday: in fact this should also be possible for the log-det stuff. Just give the matrix and all other things are initialised by default | 01:08 |
@HeikoS | lambday: If you want, you can write another entrance task on that :) | 01:08 |
@HeikoS | lambday: do you want? | 01:09 |
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has quit [Ping timeout: 244 seconds] | 01:09 | |
lambday | HeikoS: wow - so, no explictly creating linear operators | 01:09 |
lambday | that would be cool! | 01:09 |
@HeikoS | lambday: yeah, imagine you want the log-determinant of a sparse matrix | 01:09 |
@HeikoS | and just instanciate the main class with the matrix, everything else is done in the constructor to default methods | 01:10 |
@HeikoS | lambday: thats a nice entrance task, easy code, maybe some unit tests afterwards, but thats all | 01:10 |
lambday | HeikoS: alright! | 01:10 |
sunil1337 | HeikoS:,lambday: i'm up :p | 01:10 |
lambday | HeikoS: I will put up an entrance task on that! | 01:10 |
lambday | sunil1337: cool! | 01:10 |
@HeikoS | lambday: nice, thanks! | 01:10 |
-!- aroma [75ef5e6f@gateway/web/freenode/ip.117.239.94.111] has quit [] | 01:14 | |
thoralf | HeikoS: Regarding your question: IMO we can merge, since it fixes out-of-bound errors in the corresponding solvers, adds a unit test, etc. | 01:20 |
thoralf | HeikoS: But: Not all of my changes are test-backed. | 01:21 |
@HeikoS | thoralf: any chance that happens? :) | 01:21 |
thoralf | HeikoS: I'd like to, but I'm in doubt it will... | 01:22 |
@HeikoS | thoralf: is that something somebody else can do easily? | 01:22 |
@HeikoS | thoralf: like intro task? | 01:22 |
thoralf | Okay, let me check. | 01:22 |
thoralf | Basically it's cleanup (-> BuffSize), typos in includes, adding bounds checks that didn't exist before, adding a unit test *and* fixing an off-by-one that I already fixed in the other solver. | 01:24 |
shogun-notifier- | shogun: Thoralf Klein :develop * 70453d8 / src/shogun/structure/ (5 files): https://github.com/shogun-toolbox/shogun/commit/70453d8384397010416a239378d472e6373fef43 | 01:25 |
shogun-notifier- | shogun: Different simple fixes: | 01:25 |
shogun-notifier- | shogun: * Getting imports right (mixed up libp3bm.h and libppbm.h) | 01:25 |
shogun-notifier- | shogun: * Not defining BufSize in every so solver by defining it extern in libbmrm.h | 01:25 |
shogun-notifier- | shogun: * Correcting wrong initialization of beta[i+1] in libppbm.cpp | 01:25 |
shogun-notifier- | shogun: * Termination criteria: exit loop if next iteration would exceed CP buffer | 01:25 |
shogun-notifier- | shogun: Thoralf Klein :develop * a7f8ec5 / tests/unit/structure/DualLibQPBMSOSVM_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/a7f8ec53fcd4e98391ea4a31af960d3a2d23bae9 | 01:25 |
shogun-notifier- | shogun: Using gtest to run the same tests for all solvers. | 01:25 |
shogun-notifier- | shogun: * Outsourced dummy data to helper functions create_test_labels() and create_test_features(). | 01:25 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 7c6f9eb / / (6 files): https://github.com/shogun-toolbox/shogun/commit/7c6f9eb5c858c188982d27c8658a1d2d6249e151 | 01:25 |
shogun-notifier- | shogun: Merge pull request #1854 from tklein23/testing_all_bundle_method_solvers | 01:25 |
shogun-notifier- | shogun: | 01:25 |
shogun-notifier- | shogun: Testing all bundle method solvers (plus cleaning and bugfixes) | 01:25 |
thoralf | Cool. | 01:26 |
thoralf | HeikoS: First use of parametrized unit-tests: tests/unit/structure/DualLibQPBMSOSVM_unittest.cc | 01:27 |
thoralf | In case you need an example. | 01:27 |
@HeikoS | thoralf: ah sweet | 01:27 |
@HeikoS | I saw that in stan today and liked it there :) | 01:28 |
@HeikoS | thoralf: btw EXPECT is better in unit test, it doesnt terminate the program and other EXPECTS can be checked, ASSERT blocks directly, so should only be used if the program otherwise would crash (like null pointer read) | 01:28 |
thoralf | I know, but if in doubt (or too lazy or too tired) I'm using asserts. ;) | 01:29 |
@HeikoS | thoralf: lazy boy :) | 01:30 |
@HeikoS | thoralf: ok I am off home now | 01:30 |
@HeikoS | see you! | 01:30 |
thoralf | Bye. | 01:30 |
thoralf | See you all soon... | 01:30 |
-!- claughinghouse [~claughing@pool-108-28-80-98.washdc.fios.verizon.net] has joined #shogun | 01:30 | |
-!- thoralf [~thoralf@91-66-35-191-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 01:30 | |
lambday | HeikoS: https://github.com/shogun-toolbox/shogun/issues/1932 | 01:31 |
lambday | HeikoS: could you please mark it as entrance | 01:31 |
@HeikoS | lambday: done + whitespace fix :) | 01:32 |
@HeikoS | lambday: could you write an example call into this? | 01:32 |
lambday | HeikoS: example call? | 01:32 |
@HeikoS | log_est=LogDetEstimator(SGSparseMatrix m) | 01:33 |
lambday | oh yeah sure | 01:33 |
@HeikoS | like the python code that this will result in | 01:33 |
@HeikoS | lambday: nice thanks, well done, first entrance taks for you :) | 01:34 |
@HeikoS | lambday: I am going home now, see you tomorrow! | 01:34 |
lambday | HeikoS: yeah! :D | 01:34 |
lambday | HeikoS: alright! I will doze off too now - | 01:34 |
lambday | ciao tomorrow :) | 01:34 |
@HeikoS | lambday: good night! :) | 01:34 |
lambday | editing a bit :) | 01:34 |
lambday | gnyt :) | 01:34 |
-!- HeikoS [~heiko@nat-174-158.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 240 seconds] | 01:38 | |
shogun-notifier- | shogun: tklein23 :develop * 9b37dc1 / NEWS: https://github.com/shogun-toolbox/shogun/commit/9b37dc1e630d54a9c16f2d19b6a10c34d8aef73a | 01:48 |
shogun-notifier- | shogun: Update NEWS | 01:48 |
-!- lambday [67157f4e@gateway/web/freenode/ip.103.21.127.78] has quit [Quit: Leaving] | 02:02 | |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has joined #shogun | 02:25 | |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 02:43 | |
-!- bgamlath [~bgamlath@103.247.51.40] has joined #shogun | 02:49 | |
-!- bgamlath [~bgamlath@103.247.51.40] has quit [Remote host closed the connection] | 03:09 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 03:15 | |
-!- claughinghouse [~claughing@pool-108-28-80-98.washdc.fios.verizon.net] has left #shogun [] | 03:30 | |
-!- travis-ci [~travis-ci@ec2-54-197-73-137.compute-1.amazonaws.com] has joined #shogun | 03:33 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/20098068 | 03:33 |
-!- travis-ci [~travis-ci@ec2-54-197-73-137.compute-1.amazonaws.com] has left #shogun [] | 03:33 | |
-!- ameya [62c000db@gateway/web/freenode/ip.98.192.0.219] has joined #shogun | 03:46 | |
ameya | hello | 03:46 |
-!- ameya is now known as Guest40930 | 03:46 | |
-!- Saurabh7 [~Saurabh7@115.248.130.148] has quit [Ping timeout: 264 seconds] | 03:50 | |
-!- Guest40930 [62c000db@gateway/web/freenode/ip.98.192.0.219] has quit [Quit: Page closed] | 04:06 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 04:48 | |
-!- sonne|osx [~sonne@f053034067.adsl.alicedsl.de] has joined #shogun | 05:48 | |
-!- sonne|osx [~sonne@f053034067.adsl.alicedsl.de] has quit [Client Quit] | 05:49 | |
-!- cameo54321 [27293f31@gateway/web/freenode/ip.39.41.63.49] has joined #shogun | 07:53 | |
-!- cameo54321 [27293f31@gateway/web/freenode/ip.39.41.63.49] has quit [Ping timeout: 245 seconds] | 08:10 | |
-!- lisitsyn [~lisitsyn@80.252.20.67] has left #shogun [] | 08:58 | |
@besser82 | wiking: I'm back! What was your question? | 09:22 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 09:38 | |
-!- ratneshmadaan [b49531e7@gateway/web/freenode/ip.180.149.49.231] has joined #shogun | 09:52 | |
lisitsyn | ratneshmadaan: there is no particular task for deep learning I think | 09:54 |
lisitsyn | (yet) | 09:54 |
lisitsyn | and it actually doesn't matter whether you do something related or not | 09:55 |
lisitsyn | these tasks are useful to understand all the things about shogun structure, how it builds, etc | 09:55 |
ratneshmadaan | Okay, so I should follow the general guidelines then? | 09:56 |
lisitsyn | so if you see any issue you'd like to resolve just do it - this would be helpful | 09:56 |
ratneshmadaan | On it. Thank you. | 09:56 |
lisitsyn | yeah | 09:56 |
ratneshmadaan | And please update us whenever any specific task for it comes up. | 09:57 |
-!- lisitsyn is now known as 7JTAAGOXJ | 10:01 | |
-!- ratneshmadaan [b49531e7@gateway/web/freenode/ip.180.149.49.231] has quit [Ping timeout: 245 seconds] | 10:22 | |
@wiking | besser82: nevermind | 10:26 |
-!- 7JTAAGOXJ is now known as lisitsyn_ | 10:26 | |
@besser82 | wiking: allrighty :D | 10:26 |
@wiking | lisitsyn_: here? | 10:50 |
@wiking | lisitsyn_: do u know these two: http://icl.cs.utk.edu/magma/ http://icl.cs.utk.edu/plasma/index.html ? | 10:52 |
lisitsyn_ | wiking: I heard about it | 10:55 |
lisitsyn_ | wiking: actually I know a guy who used it IIRC | 10:55 |
@wiking | lisitsyn_: seems like some stuff we should slowly try to use | 10:56 |
@wiking | especially with eigen solvers... and other factorization stuff.. | 10:58 |
lisitsyn_ | wiking: indeed | 11:10 |
lisitsyn_ | wrapping such code is no rocket science :) | 11:10 |
lisitsyn_ | however I spent the best days of my life writing arpack wrapper | 11:11 |
lisitsyn_ | which had 18 parameters maybe | 11:11 |
lisitsyn_ | but still should be relatively easy to try this | 11:11 |
lisitsyn_ | wiking: the only thing is how do we employ it - do we have any particular algorithms that would gain from that? | 11:11 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 11:20 | |
shogun-notifier- | shogun: pl :develop * 49a8bcc / doc/md/README_developer.md: https://github.com/shogun-toolbox/shogun/commit/49a8bccc829efa28df3fd5258b31d8d9e4ae6803 | 11:20 |
shogun-notifier- | shogun: Unit-testing guide is too old, we use cmake instead of ./configure | 11:20 |
shogun-notifier- | shogun: pl8787 :develop * 252067a / / (76 files): https://github.com/shogun-toolbox/shogun/commit/252067aa8b1ce01f53843b9120e427eabec205ed | 11:20 |
shogun-notifier- | shogun: Merge remote-tracking branch 'shogun/develop' into develop | 11:20 |
shogun-notifier- | shogun: pl8787 :develop * d0fcc1d / doc/md/README_developer.md: https://github.com/shogun-toolbox/shogun/commit/d0fcc1d4e24bdf0deedf6153fe652801c49d716a | 11:20 |
shogun-notifier- | shogun: add more info. | 11:20 |
shogun-notifier- | shogun: pl8787 :develop * 8cb3193 / doc/md/README_developer.md: https://github.com/shogun-toolbox/shogun/commit/8cb3193c71e458375fe145971eda58409c38edff | 11:20 |
shogun-notifier- | shogun: add more info. | 11:21 |
shogun-notifier- | shogun: pl8787 :develop * 4a593a6 / tests/unit/regression/lars_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/4a593a63e7e0e9d908650524389b0de62751b2fe | 11:21 |
shogun-notifier- | shogun: Fix bug #1934. Add #ifdef HAVE_LAPACK, and can compile w/o LAPACK. | 11:21 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 441253a / doc/md/README_developer.md,tests/unit/regression/lars_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/441253a28eb653163d4583ba7126df02030e8681 | 11:21 |
shogun-notifier- | shogun: Merge pull request #1933 from pl8787/develop | 11:21 |
shogun-notifier- | shogun: | 11:21 |
shogun-notifier- | shogun: Unit-testing guide is too old, we use cmake instead of ./configure | 11:21 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 9c67564 / doc/md/README_developer.md: https://github.com/shogun-toolbox/shogun/commit/9c67564278abd5a13efe9ae016f8b3e01bf209f9 | 11:22 |
shogun-notifier- | shogun: Minor typo fixes in README_developer.md | 11:22 |
@wiking | lisitsyn_: yeah. git grep EigenSolv in ./src | 11:25 |
@wiking | lisitsyn_: that's one thing.... but then if you check qp solvers they always do some sorts of decomposition like cholensky or qr | 11:26 |
@wiking | that'd be another win | 11:26 |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 11:29 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Quit: PirosB3] | 11:40 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has joined #shogun | 12:02 | |
PirosB3 | hey guys | 12:34 |
@besser82 | PirosB3: hey! 'sup? | 12:37 |
PirosB3 | hey besser82 :) | 12:37 |
PirosB3 | I am going crazy with there OSX errors | 12:37 |
PirosB3 | So I changes SGIO.h | 12:37 |
@besser82 | PirosB3: doesn't sound too good :( | 12:37 |
PirosB3 | ahaha, it's okay. I am learning | 12:37 |
PirosB3 | I changes SGIO.h to make CONST_DIRENT_T be const struct dirent | 12:38 |
-!- prams [~prams@122.166.158.165] has joined #shogun | 12:38 | |
PirosB3 | unfortunately the build fails with: | 12:39 |
PirosB3 | cannot initialize a variable of type 'struct dirent *' with an lvalue of | 12:39 |
PirosB3 | type 'int (const struct dirent *)' | 12:39 |
PirosB3 | struct dirent* d = *SGIO::filter; | 12:39 |
PirosB3 | in features/StringFeatures | 12:39 |
PirosB3 | this is because my mac version is 10.7, and not the latests | 12:39 |
PirosB3 | if I make changes to the StringFeatures file, I get 'int (*)(const struct dirent *)' to 'int (*)(struct dirent*)' for 3rd argument | 12:41 |
PirosB3 | my idea is to assign a new variable and copy "const struct dirent", in order to make it not a constant | 12:42 |
PirosB3 | but it still is not working | 12:42 |
PirosB3 | do you have any hints? | 12:42 |
-!- CoolMcCool [quassel@fedora/besser82] has joined #shogun | 12:54 | |
-!- mode/#shogun [+o CoolMcCool] by ChanServ | 12:54 | |
-!- prams [~prams@122.166.158.165] has quit [Remote host closed the connection] | 12:56 | |
-!- besser82 [quassel@fedora/besser82] has quit [Ping timeout: 265 seconds] | 12:57 | |
-!- CoolMcCool is now known as besser82 | 12:57 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 12:59 | |
@wiking | this is crazy: https://coveralls.io/files/148275803 | 13:27 |
@wiking | only 27% of the liblinear code in the shogun codebase is being tested/used | 13:28 |
-!- makokal [~batman@kgb.informatik.uni-freiburg.de] has quit [Quit: Konversation terminated!] | 13:45 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 13:46 | |
PirosB3 | @besser82 @wiking did anyone see my proposal? I still have no answers? do you think I should enhance it? | 13:54 |
-!- lambday [67157d4e@gateway/web/freenode/ip.103.21.125.78] has joined #shogun | 14:04 | |
@besser82 | PirosB3: give it a few days to settle and have everybody read it ;) | 14:07 |
@besser82 | PirosB3: most of us still have accounts for Real Life (tm), which was the predecessor of Second Life (tm) :-P | 14:08 |
PirosB3 | ahaha ok. In the meantime I am still trying to install Shogun :) | 14:09 |
PirosB3 | I am really concerned on what I should do, I need to get it installed by today | 14:09 |
* besser82 thinks about reactivating his Real Life (tm)-account somewhen... | 14:09 | |
@besser82 | PirosB3: I'm not sure... Did you try use bundled Eigen3, yet? | 14:19 |
PirosB3 | yep :p | 14:19 |
@besser82 | PirosB3: and doesn't help... | 14:19 |
@besser82 | PirosB3: how about upgrading to some newer ver of OSX? | 14:20 |
PirosB3 | that's what I was thinking | 14:20 |
PirosB3 | I am just unsure everything will blow up :p | 14:20 |
@besser82 | PirosB3: Shouldn't that happen on Apple-devices? | 14:21 |
@besser82 | PirosB3: At least Apple *tries* to make us think it won't :-P | 14:21 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 14:22 | |
PirosB3 | ahahah | 14:25 |
PirosB3 | I could try | 14:25 |
PirosB3 | should I? or I was thinking Docker, but then... | 14:26 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 14:27 | |
lisitsyn_ | PirosB3: hey, I am currently reading your saas proposal | 14:34 |
PirosB3 | yay :) | 14:34 |
lisitsyn_ | it sounds quite cool indeed | 14:34 |
PirosB3 | what do you think of it | 14:34 |
lisitsyn_ | the only thing worries me | 14:34 |
PirosB3 | oh thanks | 14:34 |
lisitsyn_ | is that it is like a huge amount of work | 14:34 |
PirosB3 | I know, for this reason I want all the community to discuss feasible deadlines and reshape the idea if necessary | 14:35 |
lisitsyn_ | I'll think about it and write some answer later | 14:35 |
lisitsyn_ | be sure you'll get more feedback soon | 14:35 |
PirosB3 | but this could be the start of a big thing. I don't think other people have done it | 14:35 |
PirosB3 | thanks lisitsyn_ | 14:35 |
PirosB3 | it means a lot, I really feel the idea can be very helpful | 14:35 |
lisitsyn_ | there is some lag in our responses due to thing mentioned by besser82 ;) | 14:35 |
PirosB3 | let me know when u have feedback :) | 14:36 |
lisitsyn_ | PirosB3: I'll just answer your mail to the mailing list | 14:36 |
PirosB3 | sure lisitsyn_ I just wanted to make sure the mail went through, as there is no way to browse the ML (or is there?) | 14:36 |
lisitsyn_ | PirosB3: hmm actually there is | 14:36 |
lisitsyn_ | PirosB3: | 14:36 |
lisitsyn_ | http://blog.gmane.org/gmane.comp.ai.machine-learning.shogun | 14:36 |
PirosB3 | oh great | 14:36 |
PirosB3 | lisitsyn_: in the meantime, is there something specific I should be looking at? | 14:51 |
lisitsyn_ | PirosB3: sorry don't get it, what do you mean? | 14:55 |
PirosB3 | well, I want to make sure I can define a good API. Are there any particular bits of shogun internals I should be looking at for this task? lisitsyn_ | 14:55 |
lisitsyn_ | PirosB3: oh well it would need to have pretty wide perspective on what we have | 14:56 |
PirosB3 | ok | 14:56 |
-!- ashwinkachhara [67157d4c@gateway/web/freenode/ip.103.21.125.76] has joined #shogun | 15:16 | |
-!- ashwinkachhara [67157d4c@gateway/web/freenode/ip.103.21.125.76] has quit [Ping timeout: 245 seconds] | 15:20 | |
-!- ashwink [67157e4e@gateway/web/freenode/ip.103.21.126.78] has joined #shogun | 15:25 | |
-!- PirosB3 [~pirosb3@162.219.6.191] has quit [Quit: PirosB3] | 15:59 | |
sunil1337 | which opfunc can be used for complex matrices in ~/shogun/mathematics/linalg/ratapprox/logdet/opfunc or are they generic... the ipython notebook just covers reals | 16:12 |
sunil1337 | lambday: which opfunc can be used for complex matrices in ~/shogun/mathematics/linalg/ratapprox/logdet/opfunc or are they generic... the ipython notebook just covers reals | 16:15 |
lambday | sunil1337: log-determinant is only possible with real valued matrices | 16:16 |
lambday | sunil1337: just add the float64_t one :) | 16:16 |
sunil1337 | lambday: okay! | 16:17 |
sunil1337 | lambday: thanks almost done... | 16:18 |
lambday | sunil1337: good job :) if possible, add a libshogun example too :) | 16:19 |
sunil1337 | lambday: sure. i'll try.. | 16:19 |
sunil1337 | lambday: i'm using CDenseMatrixExactLog as default | 16:20 |
lambday | sunil1337: that would be the case when you're adding constructor for SGMatrix | 16:20 |
sunil1337 | lambday: yup... | 16:20 |
lambday | sunil1337: typically one wishes to use this for sparse matrices - so don't forget to add that | 16:21 |
-!- HeikoS1 [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has joined #shogun | 16:21 | |
sunil1337 | lambday: okay.... | 16:21 |
lambday | HeikoS1: hey! | 16:21 |
HeikoS1 | lambday: hi | 16:24 |
lambday | HeikoS1: trying to chalk out the draft of btest mmd class - writing a gist for confirmation - will show you in a while | 16:25 |
lambday | pseudocode | 16:25 |
HeikoS1 | lambday: great! | 16:25 |
@wiking | btw it's funny that somebody wants to do android port of shogun but not windows :S | 16:26 |
lambday | HeikoS1: oh and should I things for different number of samples for quadratic time mmd? | 16:26 |
HeikoS1 | lambday: what would be great! | 16:26 |
lambday | HeikoS1: altough for btest we need equal numbers - according to the paer | 16:26 |
lambday | HeikoS1: okies | 16:26 |
lambday | HeikoS1: oh oh - and the statistic returned is m*MMD | 16:27 |
HeikoS1 | lambday: yeah otherwise one cannot do blocks | 16:27 |
lambday | what should it be when m /= n | 16:27 |
HeikoS1 | lambday: so orient that on the linear tim eMMD implementation | 16:27 |
HeikoS1 | lambday: in fact, I think its only a few lines different since this one already streams blocks | 16:27 |
HeikoS1 | lambday: but now the blocks need to be computed in quadratic time averages | 16:27 |
lambday | HeikoS1: brb after dinner - will show the gist in a while :) | 16:30 |
HeikoS1 | lambday: enjoy your diner! | 16:30 |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 16:40 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 16:44 | |
-!- lisitsyn_ [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 16:57 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 17:00 | |
-!- AlexandraZ [bc196ac7@gateway/web/freenode/ip.188.25.106.199] has joined #shogun | 17:06 | |
-!- ashwink [67157e4e@gateway/web/freenode/ip.103.21.126.78] has left #shogun [] | 17:11 | |
-!- ashwink [67157e4e@gateway/web/freenode/ip.103.21.126.78] has joined #shogun | 17:12 | |
lambday | re | 17:15 |
-!- giri [b49531e2@gateway/web/freenode/ip.180.149.49.226] has joined #shogun | 17:18 | |
-!- giri [b49531e2@gateway/web/freenode/ip.180.149.49.226] has quit [Client Quit] | 17:20 | |
-!- cameo54321 [27296adc@gateway/web/freenode/ip.39.41.106.220] has joined #shogun | 17:36 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Quit: Page closed] | 17:43 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 17:44 | |
-!- AlexandraZ [bc196ac7@gateway/web/freenode/ip.188.25.106.199] has quit [Ping timeout: 245 seconds] | 17:45 | |
@besser82 | lambday, sunil1337: sry, for ot, but is there any special-type of curry (the spice) you can recommend? | 17:49 |
lambday | besser82: lol what're you planning to have? | 17:49 |
@besser82 | lambday: using it to spicen-up checken-breast ;) | 17:49 |
@besser82 | lambday: I love it hot btw... | 17:50 |
lambday | besser82: you're cooking? | 17:50 |
@besser82 | lambday: yes, everyday ;) | 17:50 |
lambday | besser82: lol yeah we all love breasts - :P | 17:50 |
@besser82 | lambday: hehe | 17:50 |
@besser82 | lambday: Archievement unlocked: Found besser82's special skillz | 17:51 |
sunil1337 | @besser82: well...don't really knw.. all i do is eat :p | 17:51 |
lambday | besser82: um... I know nothing about cooking :-/ may be ginger/garlic etc :-/ | 17:51 |
lambday | lol :D | 17:51 |
sunil1337 | lambday: lol | 17:51 |
@besser82 | lambday: I was more refering to the type of curry-mixture... | 17:51 |
@besser82 | lambday: Afaik there are ~80 different like this in India, e.g. Golden Elephant, ... | 17:52 |
@besser82 | sunil1337: ^^^ | 17:52 |
lambday | besser82: let me check - sunil1337: you think we'd have any luck with Sanjeev Kapoor website? | 17:52 |
sunil1337 | lambday: yeah...for sure | 17:52 |
-!- jia [a9e93511@gateway/web/freenode/ip.169.233.53.17] has joined #shogun | 17:53 | |
lambday | besser82: check this one out - http://www.sanjeevkapoor.com/ | 17:53 |
lambday | you can give your preference and it would suggest you the recipe | 17:53 |
lambday | oh shit - sorry may be not that advance :D | 17:54 |
sunil1337 | lambday: we can even try masterchef episodes :p | 17:54 |
lambday | sunil1337: never watched any :D may be mom can tell | 17:55 |
@besser82 | lambday: i need to buy some new curry-spice, but I'm tired of this yellow-orange one ;) | 17:55 |
lambday | besser82: I will ask someone who knows cooking ;) then will tell you | 17:55 |
@besser82 | lambday: I want some indian-spice which is more hot, like burning 2-times | 17:55 |
@besser82 | lambday: thx!!! | 17:55 |
sunil1337 | @besser82: try this http://www.vahrehvah.com/butter-chicken-1#.UxdWz-O1aoE | 17:55 |
lambday | sunil1337: nah butter chicken is a sweet one - he wants hot | 17:56 |
lambday | may be, chicken tikka masala or chicken kolhapuri | 17:56 |
sunil1337 | okay... | 17:56 |
@besser82 | sunil1337 lambday: yes! but there still written "Red Chilli Powder", afaik there are a f*cking lot different "Red Chilli Powders" avail... | 17:56 |
lambday | besser82: you know best - all I too do is eat and order home delivery :D | 17:57 |
@besser82 | lambday: hehe | 17:58 |
lambday | besser82: you live in Berlin, no? there must be some Indian restaurants :P | 17:58 |
-!- jia [a9e93511@gateway/web/freenode/ip.169.233.53.17] has quit [Quit: Page closed] | 17:58 | |
@besser82 | lambday: and I thought some indish ppl. might know more about that... :-P | 17:58 |
@besser82 | lambday: nope, near Hannover | 17:58 |
@besser82 | lambday: and here are just lousy one | 17:58 |
@besser82 | lambday: next good one is ~ 120 km / 80 mi | 17:59 |
lambday | ugh | 17:59 |
lambday | hmm | 17:59 |
sunil1337 | @besser82: http://www.vahrehvah.com/spicy-chicken-curry-1#.UxdX1OO1aoE try this | 17:59 |
lambday | btw I just had chicken sukkha (spicy chicken curry - thick gravy) | 18:00 |
lambday | sunil1337: yum :D | 18:00 |
@besser82 | sunil1337, lambday: that one sounds good :D | 18:01 |
sunil1337 | lambday: please i'm at HOSTEL :p | 18:01 |
@besser82 | sunil1337, lambday: tikka masala is fine, too, but had that too much :D | 18:01 |
lambday | lol we're all foodies here it seems :D | 18:01 |
cameo54321 | Haha Hi I just came here, and this seems to be a indian chefsirc channel :D | 18:03 |
cameo54321 | chefs irc* | 18:04 |
cameo54321 | Great, You guys have got appreciation for taste... :) | 18:04 |
lambday | cameo54321: bet you saved some space for all the spices :D | 18:05 |
sunil1337 | lambday: for sparse matrix do we also need to get dense? | 18:07 |
lambday | sunil1337: for sparce matrices keep the cgm one | 18:08 |
cameo54321 | :D I mean the moment I read tikka masala, I was like how did "this" topic started? | 18:08 |
sunil1337 | lambday: okay! | 18:08 |
kislay | I am having an error regarding the usage of RealFeatures on uint8 data. I am not getting past through this. :( | 18:08 |
cameo54321 | Oops You guys were converging back to work, and I added tikka masala again.. :p | 18:09 |
lambday | cameo54321: np ;) always up for food related discussions ;) | 18:09 |
lambday | kislay: what kind of error? | 18:09 |
cameo54321 | Can somoeone get me started on shogun please? | 18:10 |
lambday | real features are defined for float64_t I guess | 18:10 |
kislay | umm.. something like this : NotImplementedError: Wrong number or type of arguments for overloaded function 'new_RealFeatures'. | 18:10 |
cameo54321 | lambday Haha sure... ;) | 18:10 |
lambday | cameo54321: have you been able to make/install shogun? | 18:10 |
cameo54321 | How should I use eclipse for developing it? | 18:10 |
cameo54321 | lambday yup installed it | 18:11 |
cameo54321 | got even a python example working | 18:11 |
lambday | cameo54321: well, HeikoS1 uses eclipse - he must know better - I use vim :( | 18:11 |
lambday | cameo54321: great! :) | 18:11 |
cameo54321 | but how to move on | 18:11 |
lambday | cameo54321: there are some entrance issues on the github page | 18:11 |
HeikoS1 | lambday, cameo54321 vim is also great, in particular with clang autocomplete | 18:11 |
lambday | cameo54321: pick something and give it a shot! | 18:11 |
HeikoS1 | I am just used to using my mouse ;) | 18:12 |
HeikoS1 | but I use it only as editor and for browsing code, not for building | 18:12 |
HeikoS1 | and its slooooow | 18:12 |
lambday | HeikoS1: yeah I use clang-complete although dunno how to use all the features :P | 18:12 |
lambday | HeikoS1: yeah! | 18:12 |
lambday | HeikoS1: I guess using netbeans we can even do git related stuffs - pull/push/commit etc - not sure though :-/ | 18:13 |
cameo54321 | HeikoS1 I have to make something, and right now eclipse seems a viable option | 18:13 |
HeikoS1 | lambday: with eclipse too, but its even slower than terminal :) | 18:13 |
cameo54321 | as for vim, I'll have to learn that as well | 18:13 |
lambday | HeikoS1: yeah, love the green fonts on black background - feels like Neo :D | 18:13 |
HeikoS1 | lambday: have that too :) | 18:14 |
lambday | cameo54321: if you're starting with a new editor - you might wanna try emacs - I am just too lazy to switch but I guess it provides a lot more features | 18:14 |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has quit [Read error: Operation timed out] | 18:14 | |
lambday | cameo54321: specially while using gdb | 18:14 |
sunil1337 | cameo54321: best place to start http://www.shogun-toolbox.org/doc/en/latest/index.html | 18:14 |
cameo54321 | I am comfortable with eclipse at the moment | 18:15 |
kislay | some images when flattened and compacted in the form of row vectors are by default coming in uint8 dtype, but the RealFeature doesn't accept this datamatrix | 18:15 |
lambday | cameo54321: yeah don't worry about editors - just start hacking ;) | 18:15 |
cameo54321 | lambday exactly that is what I am looking forward to.. | 18:16 |
cameo54321 | HeikoS1 can you please help me with getting started on eclispe, with shogun? | 18:16 |
HeikoS1 | cameo54321: so all i do is to import existing code as makefile project from the src folder ,and done | 18:17 |
HeikoS1 | cameo54321: I think one can do better, but I dont really care | 18:17 |
cameo54321 | HeikoS1 and if we make some changes and build again, everything works fine? | 18:18 |
HeikoS1 | cameo54321: yeah, but run make from a terminal | 18:18 |
cameo54321 | Oh ok, this is the small point I was missing.. | 18:19 |
HeikoS1 | I use two screens so one has the terminal | 18:19 |
HeikoS1 | and for running shogun programs, I have my own little makefile, just edit them from eclipse | 18:19 |
HeikoS1 | and the unit test I edit them in gedit :) | 18:19 |
HeikoS1 | lambday: but dont do that at home, its the lazy man's way :D | 18:20 |
cameo54321 | and if I want to write a new program like the one here https://github.com/cameo54321/shogun/blob/develop/doc/md/README_developer.md | 18:20 |
HeikoS1 | just compile,link and run | 18:20 |
cameo54321 | which has its own main function | 18:20 |
lambday | HeikoS1: you use terminator for two screens? | 18:21 |
cameo54321 | HeikoS1 thanks | 18:21 |
HeikoS1 | lambday: no, I just use ubuntu ;) | 18:22 |
lambday | HeikoS1: vertically/horizontally terminal split is possible without any addons? | 18:22 |
lambday | oh I see what you meant | 18:23 |
HeikoS1 | lambday: no physical screen | 18:23 |
HeikoS1 | and tabs for terminals | 18:23 |
lambday | yeah just got that :D | 18:23 |
lambday | HeikoS1: please have a look - https://gist.github.com/lambday/9371699 | 18:24 |
lambday | HeikoS1: I am a bit confused since there are so many stuffs for which we'll just have to copy code from LinearTimeMMD | 18:24 |
lambday | HeikoS1: and I don't like multiple inheritance | 18:24 |
lambday | HeikoS1: but QuadraticTimeMMD has to be the base for BTestMMD I think | 18:24 |
HeikoS1 | lambday: no BTest is yet another two sample test | 18:25 |
HeikoS1 | lambday: or actually its closer to the linear time mmd one | 18:25 |
HeikoS1 | but I think a seperate class should be good | 18:25 |
-!- sejdino [~dino@gatsby.ucl.ac.uk] has joined #shogun | 18:25 | |
HeikoS1 | lambday: I will look at bit later as I have to finish something now | 18:25 |
HeikoS1 | but here comes sejdino | 18:25 |
HeikoS1 | he can rewiew that :) | 18:25 |
HeikoS1 | review | 18:25 |
lambday | HeikoS1: but reusing the code that quadratic time mmd already had for compute_statistic | 18:26 |
lambday | HeikoS1: alright | 18:26 |
sejdino | hey both | 18:26 |
lambday | sejdino: hi :D | 18:26 |
sejdino | how are things? | 18:26 |
HeikoS1 | sejdino: see - https://gist.github.com/lambday/9371699 | 18:26 |
HeikoS1 | lambday: just drafted that | 18:26 |
HeikoS1 | sejdino: we were just discussing if btest should in herit from quadratic or linear time mmd or none | 18:27 |
sejdino | interesting question | 18:27 |
lambday | sejdino: just the idea - since quadratic time mmd already had compute_statistic which computes it the exact same way, extending it to btest mmd can use that method | 18:27 |
sejdino | it's kind of inbetween the two | 18:27 |
HeikoS1 | sejdino: because linear time implementations effectively streams data in blocks, but the just averages the independent pairs | 18:27 |
HeikoS1 | sejdino: btest would just compute quadratic time mmd on those blocks, so its very close | 18:28 |
HeikoS1 | sejdino: but I dont know whether one can stream it, (does N need to be fixed? | 18:28 |
HeikoS1 | ) | 18:28 |
HeikoS1 | sejdino: maybe have a look into CLinearTimeMMD::compute_statistic | 18:28 |
HeikoS1 | to compare | 18:28 |
sejdino | I think you can still stream - you would just process a larger batch at once | 18:28 |
sejdino | but I agree that it's more practical to have it inherit quadratic time computation | 18:29 |
lambday | sejdino: I thought so, since the statistic is computed in the exact same way - but we're left with the option of copying a large part of the code from linear time mmd | 18:30 |
lambday | sejdino: in the draft I tried to see how to use QuadraticTimeMMD::compute_statistic inside BTestMMD::compute_statistic | 18:30 |
sejdino | maybe we can think about ways to make statistic computation a bit more modular and separate from tests? | 18:31 |
HeikoS1 | sejdino: the linear time one already streams in blocks | 18:31 |
-!- lisitsyn [~lisitsyn@80.252.20.67] has joined #shogun | 18:31 | |
HeikoS1 | sejdino: also, kernel selection is only available for linear time MMD classes, so a subclass would immediately work with that | 18:31 |
lambday | HeikoS1: ummm... but current implementation of CLinearTimeMMD::compute_statistic_and_variance does compute the linear time formula - so, we gotta rewrite that anyway | 18:32 |
lambday | unless we can separate it | 18:32 |
HeikoS1 | lambday: no I agree, but the way the data is accessed is the same | 18:32 |
lambday | HeikoS1: yeah that's the same | 18:32 |
HeikoS1 | stream block-> compute something | 18:32 |
sejdino | HeikoS1: it processes them in blocks which is great but computes only linear time statistic | 18:33 |
lambday | HeikoS1: you think we can may be have a separate base class for streaming features - which does all that | 18:33 |
lambday | then have linear/quadratic time formulas fed on in some subclasses? | 18:33 |
HeikoS1 | lambday: yeah maybe | 18:34 |
HeikoS1 | so another class streamingMMD from that linear time and Btest inherit | 18:34 |
HeikoS1 | lambday: but wait | 18:34 |
HeikoS1 | I think stiull the btest should inherit from linear time mmd | 18:34 |
HeikoS1 | it is a linear time test, the null distribution is based on that | 18:34 |
HeikoS1 | so the kernel selection things work immediately | 18:35 |
HeikoS1 | check out their code | 18:35 |
lambday | HeikoS1: yeah all same - just the formula is different :-/ | 18:35 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has quit [Ping timeout: 245 seconds] | 18:35 | |
HeikoS1 | lambday: ok what about pulling the streaming code into a base class then? | 18:35 |
HeikoS1 | or copy it | 18:35 |
lambday | HeikoS1: including the multiple kernels part? | 18:35 |
HeikoS1 | lambday: ah that | 18:36 |
HeikoS1 | lambday: yeah that also works for the btest | 18:36 |
lambday | yep! | 18:36 |
sejdino | HeikoS1: do they actually do any kernel selection in the paper? | 18:36 |
HeikoS1 | so why not have CStreamingMMD->CLinearTimeMMD->CBTestMMD ? | 18:36 |
HeikoS1 | sejdino: yeah they do | 18:36 |
HeikoS1 | lambday: and the streaming feature blocks code is in CStreamingMMD | 18:37 |
lambday | HeikoS1: and copy the code from QuadraticTimeMMD::compute_statistic to BTestMMD? | 18:37 |
HeikoS1 | lambday: or you leave things in CLinearTimeMMD and call an abstract method in the compute statistic code | 18:37 |
HeikoS1 | which then is overloaded | 18:37 |
HeikoS1 | lambday: I think that might be best in fact | 18:37 |
lambday | HeikoS1: yeah that's what I was thinkin' | 18:37 |
lambday | oh wait wait - | 18:37 |
HeikoS1 | lambday: the code that updates the statistic is in an overloaded method and BTest just modifies that to compute the quadratic statistic | 18:38 |
lambday | HeikoS1: ummmm - still copying code :( | 18:38 |
HeikoS1 | lambday: no, just have an instance of CQuadraticTimeMMD in the BTest | 18:38 |
HeikoS1 | as a member | 18:38 |
HeikoS1 | and then pass the block to it | 18:39 |
lambday | HeikoS1: ah that works - might look a bit ugly but works | 18:39 |
sejdino | HeikoS1: that seems like a nice solution | 18:40 |
sejdino | we might think about a different name for quadratic time mmd to make things less confusing | 18:40 |
HeikoS1 | sejdino: like ? | 18:41 |
sejdino | it's basically just a UStatistic or VStatistic MMD | 18:41 |
sejdino | which one are you using? | 18:41 |
-!- khalednasr [c4dd5cc1@gateway/web/freenode/ip.196.221.92.193] has joined #shogun | 18:41 | |
sejdino | biased or unbiased? | 18:41 |
lambday | sejdino: both are implemented | 18:42 |
lambday | sejdino: default is unbiased | 18:42 |
HeikoS1 | sejdino: both possible | 18:43 |
HeikoS1 | sejdino: but the point is that quadratic uses all pairs | 18:43 |
HeikoS1 | so quadratic is a good name isnt it? | 18:43 |
HeikoS1 | and linear just computes averages of independent things | 18:43 |
HeikoS1 | and b test does that too, but those independent things are blocks | 18:43 |
sejdino | I see, so you can basically in each of the B tests call either U or V? | 18:44 |
HeikoS1 | so its actually a linear time statistic (on blocks) | 18:44 |
HeikoS1 | sejdino: yeah | 18:44 |
sejdino | well, yes, unless you're using two blocks :) | 18:44 |
sejdino | then it becomes quadratic time | 18:44 |
sejdino | but then the kernel selection doesn't work | 18:46 |
HeikoS1 | sejdino: so what do you suggest? | 18:47 |
sejdino | HeikoS1: well, we could let the number of blocks be specified freely | 18:48 |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has joined #shogun | 18:48 | |
sejdino | but then we need to worry about the asymptotics not being valid | 18:48 |
sejdino | if the number of blocks is very small compared to the number of samples | 18:49 |
sejdino | I think it's better not to allow specifying the number of blocks but actually to specify the number of samples within a block? | 18:49 |
HeikoS1 | sejdino: yeah sampler per block is better, in particular if we are stremaing | 18:50 |
HeikoS1 | streaming | 18:50 |
sejdino | exactly | 18:50 |
HeikoS1 | sejdino: and then always do independent things as for linear time mmd. so both classes immediately can be plugged into the kernel selection classes | 18:50 |
lisitsyn | hey guys | 18:50 |
HeikoS1 | lisitsyn: hi! | 18:51 |
HeikoS1 | sejdino: meet sergey, another core shogun guy | 18:51 |
HeikoS1 | from Russia | 18:51 |
lisitsyn | glad to see discussion on this crazy project :D | 18:51 |
sejdino | Hi Sergey! | 18:51 |
HeikoS1 | lisitsyn: this is dino, postdoc at the Gatsby | 18:51 |
sejdino | nice to meet you | 18:51 |
lisitsyn | sejdino: hey | 18:51 |
lisitsyn | my pleasure ;) | 18:51 |
sejdino | likewise - very happy to be involved in shogun :) | 18:51 |
lisitsyn | sejdino: you seem to be slavic as well ;) | 18:52 |
sejdino | indeed - even 1/4 russian :) | 18:52 |
lisitsyn | moar russians for shogun! | 18:52 |
sejdino | 5/4? | 18:52 |
lisitsyn | currently yeah | 18:52 |
HeikoS1 | where is roman? :) | 18:52 |
HeikoS1 | (hes from siberia) | 18:52 |
lisitsyn | HeikoS1: not anymore | 18:53 |
HeikoS1 | st petersburg | 18:53 |
lisitsyn | now he is st.petersburg guy :D | 18:53 |
lambday | sejdino: HeikoS1: so specify the block-size, as they do in their matlab implementation | 18:54 |
HeikoS1 | lambday: yeah! | 18:54 |
sejdino | lambday: yes, that sounds good | 18:55 |
HeikoS1 | lambday: so people can decide how many points they want to stream | 18:55 |
HeikoS1 | lambday: make sure b<n ;) | 18:55 |
sejdino | I would say even b<0.1n | 18:55 |
lambday | HeikoS1: yeah :D in the matlab code they assert (blocksize >= 2 && blocksize < m / 3); | 18:55 |
sejdino | for the kernel selection part | 18:56 |
sejdino | but I guess we can allow even 2 blocks just to run a test | 18:56 |
lisitsyn | lambday: are you going to apply for this task? | 18:56 |
sejdino | because permutation test will still give correct p-value | 18:56 |
HeikoS1 | sejdino: agreed! | 18:57 |
lambday | sejdino: alright | 18:57 |
lambday | lisitsyn: yeah :D | 18:57 |
lisitsyn | lambday: I see :) | 18:57 |
HeikoS1 | sejdino, lambday it will be interesting to reproduce the paper results in the notebook, i.e. compare gaussian approximation to samples from the null via permutation | 18:58 |
sejdino | lisitsyn: we've had some excellent progress on the project already - lambday has been driving it :) | 18:58 |
lisitsyn | sejdino: that's the ideal thing | 18:58 |
lisitsyn | that's how HeikoS1 joined shogun | 18:58 |
lisitsyn | well and I did something like that | 18:58 |
sejdino | HeikoS1: I agree absolutely - it would be very nice to plot and see how it goes to a Gaussian as the number of blocks increases | 18:59 |
HeikoS1 | lambday: see the plots in the paper | 19:00 |
HeikoS1 | sejdino: do you want to write an intro task on reproducing that? | 19:00 |
sejdino | HeikoS1: it will move shape from chi-squares to a gaussian | 19:00 |
sejdino | HeikoS1: yes, I can write a task | 19:00 |
HeikoS1 | sejdino: cool! | 19:00 |
lambday | ah the plot is nice! | 19:03 |
-!- dhruv13 [~dhruv13@84.64.12.151] has joined #shogun | 19:03 | |
kislay | lambday, please if you could point me to some hack!! I am stuck for quite a while | 19:04 |
lambday | sejdino: HeikoS1: alright - I am making sure if technically things are alright with the proposed implementation | 19:08 |
lambday | kislay: sure - real features is I guess defined with float64_t - so, it would be incompatible with uint8 I guess | 19:09 |
lambday | kislay: let me make sure | 19:09 |
-!- dhruv13 [~dhruv13@84.64.12.151] has quit [Ping timeout: 264 seconds] | 19:09 | |
HeikoS1 | lambday: great! | 19:10 |
-!- cameo54321 [27296adc@gateway/web/freenode/ip.39.41.106.220] has quit [Ping timeout: 245 seconds] | 19:13 | |
-!- cameo54321 [27296adc@gateway/web/freenode/ip.39.41.106.220] has joined #shogun | 19:16 | |
kislay | yup!!. Till now i couldn't find a single way to run uint8 in RealFeatures! But,then how to convert uint8 data type to float64. | 19:24 |
-!- khalednasr [c4dd5cc1@gateway/web/freenode/ip.196.221.92.193] has quit [Quit: Page closed] | 19:25 | |
kislay | using array(Image.open(filenames[0]).convert('L'),dtype=float64) just doesn't work | 19:25 |
-!- dhruv13 [~dhruv13@84.64.12.151] has joined #shogun | 19:27 | |
lambday | kislay: I am not so sure - may be use ByteFeatures which is defined for uint8 | 19:30 |
lambday | kislay: please have a look at interfaces/modular/Features.i typemaps | 19:30 |
-!- kislay [~kislay@93.114.45.13] has quit [Read error: Connection reset by peer] | 19:31 | |
lambday | brb | 19:31 |
-!- cameo54321 [27296adc@gateway/web/freenode/ip.39.41.106.220] has quit [Ping timeout: 245 seconds] | 19:50 | |
-!- kislay [~kislay@93.114.45.13] has joined #shogun | 19:53 | |
-!- parijat [75c7d731@gateway/web/freenode/ip.117.199.215.49] has joined #shogun | 20:00 | |
-!- parijat [75c7d731@gateway/web/freenode/ip.117.199.215.49] has quit [Client Quit] | 20:02 | |
-!- sejdino [~dino@gatsby.ucl.ac.uk] has left #shogun [] | 20:24 | |
-!- HeikoS1 [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has left #shogun [] | 20:24 | |
sunil1337 | lambday: you up? | 20:27 |
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has joined #shogun | 20:42 | |
-!- sunil1337 [0e8bd057@gateway/web/freenode/ip.14.139.208.87] has left #shogun [] | 20:46 | |
-!- AlexandraZ [bc196ac7@gateway/web/freenode/ip.188.25.106.199] has joined #shogun | 21:33 | |
-!- dhruv13 [~dhruv13@84.64.12.151] has left #shogun ["Leaving"] | 22:21 | |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 22:35 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 22:35 | |
@iglesiasg | hey guys | 22:36 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 22:42 | |
shogun-notifier- | shogun: Parijat Mazumdar :develop * 10aa692 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/10aa6926652d6c581ef7ac290ceee5a638570dd7 | 22:42 |
shogun-notifier- | shogun: prevents travis failure when mldata is in flaky state | 22:42 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 2db71a6 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/2db71a62df313cd531857a8b482ccacc333c52d4 | 22:42 |
shogun-notifier- | shogun: Merge pull request #1938 from mazumdarparijat/pca | 22:42 |
shogun-notifier- | shogun: | 22:42 |
shogun-notifier- | shogun: prevents travis failure when mldata is in flaky state | 22:42 |
shogun-notifier- | shogun: Fernando Iglesias :develop * d590c99 / NEWS: https://github.com/shogun-toolbox/shogun/commit/d590c99c203092f5041af73f2922e0397de28241 | 22:46 |
shogun-notifier- | shogun: Update NEWS with libary_mldatahdf5.cpp fix. | 22:46 |
-!- iglesiasg is now known as iglesiasg_away | 23:07 | |
-!- ashwink [67157e4e@gateway/web/freenode/ip.103.21.126.78] has quit [Ping timeout: 245 seconds] | 23:16 | |
-!- iglesiasg_away [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving] | 23:36 | |
--- Log closed Thu Mar 06 00:00:26 2014 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!