IRC logs of #shogun for Friday, 2011-06-24

--- Log opened Fri Jun 24 00:00:16 2011
@sonney2kmlsec, blackburn was asking who mlsec is - that's it.00:00
@mlsecthx for the info. gotta slepp. 120% admin load each day at tu00:02
@sonney2kmlsec, only 100% play load at daytime here00:02
@sonney2kmlsec, see you soon old buddy! ETA 4.7!00:03
f-xsonney2k: i still have to call a StreamingFeatures::function to set the vector readers from within StreamingSimpleFeatures right?00:28
f-xbecause i can't call set_read_vector directly from the StreamingFeatures constructor00:28
@sonney2kf-x, why not?00:28
f-xbecause it is defined in StreamingSimpleFeatures.. the linker says my constructor cannot call a pure virtual function00:29
@sonney2kf-x, hmmhh maybe that is a c++ limitation. if you don't make it pure virtual but add print statements in both classes - will the derived one be called?00:31
f-xsonney2k: nope.. i don't think so00:31
f-xbecause during construction the base class knows only about its own members, i guess00:32
@sonney2kf-x, indeed - explanation here
f-xsonney2k: yeah.. exactly that..00:32
f-xso i can define a function say set_read_functions() in StreamingFeatures, which calls set_vector_reader and set_vector_and_label_reader of the derived class00:33
f-xand call set_read_functions from streamingsimplefeatures00:33
@sonney2kdangerous problem...00:33
@sonney2kf-x, yes00:34
f-xor some different version of init() itself, where we may later add more things00:34
f-xas we generalize things more00:34
f-xokay, putting it under 'set_read_functions' for the time being00:36
f-xsonney2k: req sent00:40
CIA-32shogun: Soeren Sonnenburg master * rec1eb8e / (7 files in 2 dirs): Merge branch 'streaming' of git:// (+6 more commits...) -
f-xsonney2k: you're welcome :)00:51
@sonney2kf-x, now finish these features and get sgd done and feel happy :)00:52
@sonney2know == next steps00:52
f-xsonney2k: yeah.. StringFeatures should be ready now00:53
f-xbut there's one thing:00:53
f-xother than the basic get_* stuff, i'm including the remap thing00:53
f-xthe remapping is done (if desired) during each call to get_next_example()00:54
f-xsay remap is from alpha1->alpha2, how do we define 'alphabet' for the object at each step?00:54
f-xshould i be clearer? i guess i'm being vague...00:55
@sonney2kalphabet needs to be set once beforehand (in constructor for example)00:56
f-xyes, alpha100:56
f-xsay i need to remap from DNA to RAWDNA00:57
f-xthe alphabet set will be DNA00:57
f-xand after every call to get_next_example, the histogram of the alphabet will be updated00:57
@sonney2k(or every char in example will be checkt00:59
f-xand if i want to remap, there will be the RAWDNA alphabet whose histogram will be updated at the same time00:59
@sonney2kno stop00:59
@sonney2kthe remap is not that flexible00:59
@sonney2kit maps only from alpha -> raw00:59
@sonney2kone cannot tranfer DNA to digits for example01:00
@sonney2kor vice versa01:00
@sonney2kso it is just 'A' -> 0 etc01:00
f-xoh.. okk01:01
f-xwell i think things are becoming clearer as i ask.. is there any other function you want added in StreamingStringFeatures?01:02
@sonney2kas I said only some get_string of the raw string :)01:05
@sonney2kwith optional translation / error checking01:05
f-xconverted to the required T type char-by-char?01:06
@sonney2kf-x, for T types other than char like e.g. int it should read binary ints...01:06
@sonney2knot a priority though now01:06
@sonney2kI would just do it for chars for now01:07
@sonney2kall the extras can come later01:07
f-xokay, so just returning the raw string will do for now01:07
-!- f-x [~user@] has quit [Remote host closed the connection]01:33
CIA-32shogun: Soeren Sonnenburg master * r16c3af6 / src/modular/Features.i : import preprocessors in features too -
-!- blackburn [~blackburn@] has joined #shogun10:37
blackburnsonney2k: hey I didn't10:42
blackburn@sonney2k ml*sec, blackburn was asking who mlsec is - that's it.10:42
blackburnah I wrote his nickname, sorry Konrad :)10:54
-!- sploving [~root@] has joined #shogun11:03
blackburnsploving: yes?11:06
splovingwhen I test python, like python *.py, it shows could not find module shogun.Features11:07
splovingblackburn, do you know how to set the lib path?11:07
blackburnsploving: nope.. it fails with python_modular interface configured?11:08
splovingblackburn, I use ./configure --interfaces=libshogun,python_modular11:08
splovingit works11:08
splovingthen "make"11:08
splovingdo i need make install?11:09
blackburnof course11:09
splovingblackburn, in java, it does not need11:10
splovingblackburn, it works!11:10
splovingit disturb me a long time11:11
splovingthen works most of "OK"11:12
splovingI am glad to see the result!11:12
splovingbut some error11:13
blackburnsploving: it is ok, we broke many things with ST* -> SGVector transition11:16
blackburnthree days ago there were ERROR for almost every test with some OK11:17 OK11:17
splovingblackburn, does that mean I install ATLAS well?11:18
blackburnsploving: yes, it works only with atlas11:18
splovingblackburn, but when ./configure,  it shows ATLAS no.11:18
splovingi think the configure need fix11:19
blackburnwell try to python preprocessor_locallylinearembedding_modular.py11:19
blackburnit exactly requires atlas, I wrote it :)11:19
splovingit print "locallylinearembedding"11:20
splovingdoes that mean well?11:20
blackburnit will fail if there was no atlas11:21
splovingblackburn, now I try to play with generator.py11:21
splovingI need write a similar java such as generator.java11:22
blackburnI see11:22
splovingwhat is used for?11:24
splovingdo I need run it before run blackburn11:24
blackburnsploving: I don't know yet11:25
blackburnI never used it before11:25
splovingblackburn, you never used testsuite?11:26
blackburnyes :D11:26
blackburnsploving: this testsuite is just some kind of 'wrapper' for test in examples11:27
blackburnbtw it is not in actual state now11:28
splovingblackburn, you mean it does not work now?11:29
blackburnyes, some things to be fixed here11:32
splovingblackburn, No module named PreProc11:37
splovingwhen I test_all.sh11:37
blackburnit should be Preprocessor11:37
splovingSo these examples need updated?11:38
blackburnI think it is in testsuite sources11:41
blackburnexamples are ok11:41
blackburnsomething is wrong with serialization11:44
splovingit is in test_one.py11:44
splovingimport preproc should be import preprocessor11:45
splovingno it is in Kernel.py11:46
splovingit said "from shogun.PreProc import *"11:47
blackburnsploving: i'm fixing it now11:48
splovingone is the other is test_one.py11:48
blackburnsploving: I guess we need some supersonney here11:55
splovingI agree11:56
blackburnI will commit fixes for get it working but not everything is working now11:56
splovingthere always some error: not fond Preproc11:56
CIA-32shogun: Sergey Lisitsyn master * r8de57a2 / (8 files in 2 dirs): Fixes for python testsuite -
blackburnsploving: should be ok now12:00
splovingI will try.12:00
blackburnnow tests fails, not testsuite :D12:01
splovingyou should fix consider two sides not just onw12:01
splovingI gtg bye blackburn12:03
splovinglooking forward to your news later!12:04
-!- sploving [~root@] has left #shogun []12:04
@sonney2kblackburn, around?12:34
@sonney2kblackburn, I have a problem doing the dotfeatures T** transition12:34
@sonney2kit fails in gaussiannaivebayes now12:35
@sonney2kand I am not sure how to properly fix it12:35
@sonney2kshould I just comment this part and commit and you do the fix?12:35
-!- blackburn [~blackburn@] has quit [Read error: Connection reset by peer]12:36
-!- blackburn [~blackburn@] has joined #shogun12:36
blackburnsorry got disconnected12:36
@sonney2k<sonney2k> blackburn, I have a problem doing the dotfeatures T** transition12:37
@sonney2k<sonney2k> it fails in gaussiannaivebayes now12:37
@sonney2k<sonney2k> and I am not sure how to properly fix it12:37
@sonney2k<sonney2k> should I just comment this part and commit and you do the fix?12:37
blackburnsonney2k: I will fix it in a minute12:37
@sonney2kblackburn, wait you cannot12:37
@sonney2kI will comment and commit such that you know12:37
blackburnwell there are some things to be fixed here12:38
blackburnactually it is for simple features12:38
blackburnsonney2k: you are talking about if at line 54 GaussianNaiveBayes.cpp, rgiht?12:39
@sonney2kblackburn, please wait12:39
@sonney2kalesis-novik, I need to touch GMM too12:43
@sonney2kand Gaussian12:43
@sonney2kblackburn, btw I still like alesis-novik idea of having that do free flag set appropriately at all times quite a lot :)12:46
blackburnsonney2k: I have no idea about that so just do it like you want :)12:50
blackburnsonney2k: ah I didn't tell you: new LMDS is really gorgeous12:54
CIA-32shogun: Soeren Sonnenburg master * r9d78a61 / (8 files in 6 dirs): T** transition for DotFeatures' get_feature_vector and matrix -
CIA-32shogun: Soeren Sonnenburg master * r9e02b38 / (8 files in 2 dirs): Merge branch 'master' of -
blackburn15000 3d vectors was embedded to 2d space with 13s taken12:54
@sonney2kblackburn, :)12:55
blackburnwith arpack and when I will write it better it could be even faster with factor of ~1.512:55
@sonney2kvery nice12:56
blackburnactually I don't know why we need ClassicMDS - it will never finish on 15K examples12:58
blackburnjust like I will never finish reading Alberts' 1725 pages Molecular Biology of The Cell book13:01
blackburnsonney2k: hey it is not compiling now13:11
blackburnDistance_wrap.cxx: In function ‘void* _p_shogun__CPruneVarSubMeanTo_p_shogun__CSGObject(void*, int*)’:13:11
blackburnDistance_wrap.cxx:55762: error: ‘CPruneVarSubMean’ is not a member of ‘shogun’13:11
blackburnand so on13:11
blackburnfor every preprocessor13:11
CIA-32shogun: Sergey Lisitsyn master * rdd0ff30 / (2 files): Removed ignored = operator at RandomFourierGaussPreproc -
blackburnsonney2k: !14:01
CIA-32shogun: Sergey Lisitsyn master * r1b78b3a / (2 files): Fixes for GaussianNaiveBayes -
CIA-32shogun: Sergey Lisitsyn master * r88edc41 / (7 files in 2 dirs): Added LandmarkIsomap and fixed Isomap and LandmarkMDS -
@sonney2kblackburn, no idea14:29
@sonney2kblackburn, seems like I can fix it by adding %include's in the modular if's14:34
@sonney2kno idea why it is suddenly there though14:34
@sonney2kI didn't do any changes there - at least I am not aware of any...14:34
blackburnme too :(14:40
blackburnsonney2k: strange thing about it is they all are including Features_includes.i14:42
blackburnand preprocessors are included there14:43
CIA-32shogun: Soeren Sonnenburg master * r5a32d4d / src/libshogun/kernel/Kernel.cpp : add missing cases to kernels -
CIA-32shogun: Soeren Sonnenburg master * re4b52f0 / (8 files): include preprocs in all modular interfaces -
@sonney2kblackburn, well swig does not include recursively14:46
CIA-32shogun: Soeren Sonnenburg master * ref42f49 / (11 files in 3 dirs): Merge branch 'master' of -
@sonney2kblackburn, it can with  the includeall directive14:46
@sonney2kbut this generates other problems I wasn't able to resolve14:46
blackburnI don't understand why it was emerged14:47
@sonney2kanyway I put the include for preprocs everywhere and now it compiles14:47
@sonney2kme neither14:47
@sonney2kme fiddling with dotfeatures* certainly shouldn't14:47
blackburnsonney2k: I guess this caused14:48
@sonney2kblackburn, nope14:48
@sonney2kI removed that and it still didn't compile here14:48
blackburnwell in this case I completely have no idea14:49
@sonney2kI even did git clean again to check14:49
@sonney2kanyway we need that include there for splovings stuff14:49
blackburnsonney2k: btw sploving have commit rights too now?14:50
@sonney2kblackburn, no14:50
blackburnI was wondering wtih14:50
blackburn sploving merged commit 16c3af6 into shogun-toolbox:master from sploving:master about 5 hours ago14:50
@sonney2kblackburn, he is probably not using rebase14:52
blackburnI see14:53
blackburnI need to invent (n^2) dijkstra analogue heheh14:58
blackburn*not dijkstra, floyd-warshall14:58
@sonney2kseems like sparse features work nicely now15:02
@sonney2kblackburn, the only big issue I have no idea why it appeared is to get the multi-class classifiers back to work15:02
@sonney2kI really wonder why the apply() function is not visible...15:03
blackburnyes, I remind15:03
blackburnlapack was not working15:03
@sonney2kblackburn, btw sploving still has no lapack/atlas15:03
blackburnsonney2k: he was running tests today15:04
@sonney2khis atlas is not compiled with -lm and libs are missing15:04
blackburnlda was working15:04
blackburnand my LLE too15:04
@sonney2knot tests - examples and yeah these fail gracefully IIRC when not installed15:04
blackburnit won't even import without lapack15:04
@sonney2kare you sure?15:04
blackburnexamples, yes15:04
-!- in3xes_ [~in3xes@] has joined #shogun15:04
blackburnsonney2k: yes, LDA is not working without lapack15:05
blackburnand I'm sure LLE too because it uses dsymv :)15:05
@sonney2kblackburn, I know not working - but the check* stuff just runs or?15:05
blackburnin test for LLE lapack routines will be called exactly15:06
blackburnI don't know how can it say there are no atlas and works at the same time15:06
-!- in3xes [~in3xes@] has quit [Ping timeout: 276 seconds]15:07
@sonney2kblackburn, works != python doesn't crash15:07
@sonney2k*sigh* java modular does no longer compile - wtf?!15:07
blackburnno idea :)15:07
@sonney2kI wish github had some build bot15:08
blackburnsonney2k: won't it crash if there are no lapack and dsymv routine is called?!15:08
@sonney2kblackburn, it is all ifdef'd remember?15:08
blackburnsonney2k: all class LocallyLinearEmbedding is ifdef'd15:09
blackburnit won't be imported15:09
@sonney2kblackburn, I mean just try15:09
blackburnah ok15:09
@sonney2kblackburn, his configure with --enable-lapack failed for a good reason15:09
@sonney2kat least15:09
blackburnsonney2k: compiling.. I believe it will fail :)15:14
-!- in3xes_ [~in3xes@] has quit [Ping timeout: 240 seconds]15:14
blackburnon import15:14
-!- in3xes_ [~in3xes@] has joined #shogun15:14
@sonney2kyeah it should15:16
@sonney2kbut - will that just say OK?15:17
blackburnshould be error I guess15:17
-!- in3xes_ is now known as in3xes15:19
@sonney2kblackburn, well let me know :D15:20
* sonney2k is f*** with this java crap15:20
blackburnoh damn tropic tolyatti hehe rainy-sunny-rainy-sunny-rainy-sunny15:23
blackburnrunning .. ERROR15:24
@sonney2kblackburn, samer here15:24
@sonney2kweather wise15:25
blackburnstrange summer15:25
blackburnlast summer was so warm that 30% of forest there burned15:25
blackburnsonney2k: my preprocessors fails, lda too15:26
@sonney2kblackburn, then I have no idea what sploving sent15:26
@sonney2khis configure failed badly15:26
blackburnsonney2k: this morning he tried and it was working on his machine15:26
@sonney2kwith all sorts of errors15:26
@sonney2kand with these it just cannot work15:27
@sonney2kno idea15:27
blackburnand he wants me to fix testsuite hehe15:27
blackburn(02:01:03 PM) blackburn: now tests fails, not testsuite :D15:27
blackburn(02:01:15 PM) sploving: why?15:27
blackburn(02:01:45 PM) sploving: you should fix consider two sides not just onw15:27
@sonney2kI cannot parse this language15:28
blackburnI'm feared he will abuse me if I won't fix tests :(15:28
@sonney2kjava is really broken now - and I don't have the slightest clue why15:30
blackburnin fact shogun is really broken now :D15:30
@sonney2kblackburn, python_modular is not or?15:31
blackburnit compiles yes15:31
blackburnlibshogunui too15:31
blackburnI mean we will fix it for a long :)15:31
blackburnwith ST* transition and other things15:31
@sonney2kblackburn, it not only compiles but almost all examples run15:32
blackburnwell are you sure examples spread over all the functionality?15:32
@sonney2kblackburn, yes15:33
blackburnhmm then it is ok15:34
-!- in3xes [~in3xes@] has quit [Ping timeout: 240 seconds]15:35
@sonney2kblackburn, well no - we need to get the examples running first15:35
@sonney2kthen run the actual test suite15:35
@sonney2kif that passes - with valgrind running and no memmory leaks etc - then it is all good15:35
blackburnsonney2k: seems only multiclass SVMs is failing now15:35
-!- in3xes [~in3xes@] has joined #shogun15:36
@sonney2kblackburn, and we don't know why it fails15:36
@sonney2kclassify is in kernel machine15:36
@sonney2kerr apply15:36
@sonney2kbut it for some reason is not seen in the derived class15:36
@sonney2khere also the serialization tests fail15:37
blackburnserialization is not working15:37
@sonney2kbecause of add_preproc :)15:37
blackburnthis morning I fixed some lines at testsuite15:38
@sonney2kahh and because of multiclass15:38
blackburntried to run, and there was some serialization issues15:38
@sonney2k(it uses GMNPSVM -> multiclass)15:38
blackburnfound in some work's references:15:40
blackburnAnonymous Author. Anonymous title15:40
@sonney2kblackburn, serialization for labels is broken15:40
blackburnwhy it is? ST*?15:40
@sonney2kblackburn, I guess heikos subset patch15:40
* sonney2k checks15:41
blackburnin the case won't try to fix it - don't understand what heiko is doing :D15:41
@sonney2kblackburn, unfortunately heiko's computer died15:41
* sonney2k is looking at it15:41
-!- heiko [865b7525@gateway/web/freenode/ip.] has joined #shogun15:41
blackburnjust in time, Heiko :)15:42
heiko :)15:42
heikowhats up?15:42
blackburnheiko: we just noticed that labels serialization is not working15:42
heikooh, ok15:42
blackburnheiko: what's with your computer?15:43
heikoI will have a look on this15:43
@sonney2kI see what the problem is15:43
@sonney2kheiko, serialization support in labels is broken15:43
@sonney2kheiko, I guess due to the num_labels_total / num_labels thingy15:43
heikowell, its old and the harddisc had an eror in the partition table or something, I am currently re-installing15:43
heikoI will check it as soon as possible15:44
@sonney2kheiko, shouldn't m_parameters->add_vector(&labels, &num_labels_total, "labels"15:44
@sonney2kerr stop15:44
@sonney2kI think we should also serialzie num_labels then - or is that no longer needed (known from subset?!)15:45
blackburnheiko: better not use old hard drives - check it's S.M.A.R.T for bad blocks15:45
blackburnif <>0 - it will increase for sure, better buy new :)15:46
heikowell yes I am planing to buy a new laptop some day when I have money ;)15:46
heikothe num_abels thing15:47
heikoI think num_labels_total has to be used since we want to serialize the complete array15:47
blackburnI guess you will have some at 17-19th of July :D15:47
@sonney2kheiko, agreed15:47
@sonney2kbut what about the other variables later?15:47
-!- in3xes [~in3xes@] has quit [Ping timeout: 240 seconds]15:47
@sonney2kthey are not serialized and so all 0 when labels is loaded15:47
@sonney2ke.g. num_labels = 0 then15:48
heikoblackburn, yes :) but I will have to keep the money to pay study fees for my master15:48
heikosonney2k, yes you are right that is the poblem15:48
blackburnheiko: I see, well just buy new hard drive for your laptop - it is cheap15:48
heikoblackburn, yes, I will think of it :)15:48
@sonney2kheiko, can't we do this via the subset classsss15:49
blackburnlosing data is FFUUUU15:49
@sonney2kI mean never having two num_labels things?15:49
heikosonney2k, yes, the subset class serializes its array and size anyways15:49
@sonney2kheiko, so I can remove num_labels right?15:49
heikoyou mean, only have one variable for the size?15:50
heikoah ok, yes this is possible15:50
@sonney2kbecause subset will give the subset size15:50
heikosome methods might have to be changed then but I think this might work15:50
@sonney2kok I am checking for num_labels* stuff15:50
heikoblackburn, I do regular backups :)15:50
heikook, alright15:50
heikoperhaps remove num_labels_total15:51
heikobecause this was the one that was added by me15:51
heikosame thing for string features by the way15:51
heikodo we have somehing like tests for serialization?15:52
alesis-noviksonney2k, what do you need to do with Gaussian and GMM?15:52
@sonney2kalesis-novik, I did already - look at the diff15:52
@sonney2ksome SGVector changes15:52
@sonney2kheiko, I think you need to fix the subset business for get labels15:55
@sonney2kI will just make this test work without caring too much about subsets15:55
heikook, I will have a look at all the suset stuff15:56
alesis-noviksonney2k, so what would happen if I made a pull request now?15:56
@sonney2kI'd like to see all tests pass for python_modular now15:56
@sonney2kalesis-novik, make sure you do git pull --rebase15:57
-!- blackburn [~blackburn@] has quit [Quit: Leaving.]16:01
-!- blackburn [~blackburn@] has joined #shogun16:01
-!- blackburn [~blackburn@] has quit [Client Quit]16:01
-!- in3xes [~in3xes@] has joined #shogun16:01
CIA-32shogun: Soeren Sonnenburg master * r8ca4620 / (4 files in 3 dirs): fix the serialization examples and Labels' serialization -
@sonney2kheiko, I think you only have to modify get_num_labels() - I would write labels like this: if you have a subset object set it uses this number / subsetting16:13
@sonney2kif not then it uses things directly16:13
@sonney2kthe other option is to always have a subset object that is just the whole set - that would work too but might be confusing16:14
heikoI would prefer the first version16:14
@sonney2kme too16:15
@sonney2kin addition these confidence things in labels can go16:15
@sonney2kno one uses these so far so no need to make things overly complex16:15
heikook, alright16:15
alesis-noviksonney2k, I hope you are up for some commenting on formatting and other problems :)16:18
@sonney2kalesis-novik, totally totally ;D16:19
@sonney2kall examples run through (python_modular) again :))16:19
@sonney2kheiko, btw please use SG_UNREF() / SG_REF for the subset - this way you can even just return the whole subset object16:41
@sonney2kno need for wrappers then16:41
-!- f-x [~user@] has joined #shogun16:53
@sonney2kalesis-novik, a monster patch!16:53
@sonney2kit is soo big that I cannot really check it16:54
@sonney2kalesis-novik, I will just merge - do the minor change please17:11
@sonney2kbtw it would be great if you could fully use SGVector & friends17:11
@sonney2kinstead of these ptr / size thingies17:11
@sonney2kthen for serialization you can just call m_parameters->add(&vec) etc17:12
CIA-32shogun: Soeren Sonnenburg master * r8ace2c2 / (6 files in 3 dirs): Merge branch 'gmm' of git:// (+8 more commits...) -
alesis-noviksonney2k, that's the plan.17:13
@sonney2kok great17:13
@sonney2kI've merged as is - but please do the overhaul then :)17:13
alesis-novikand the n_ thing was because I saw it somewhere else :)17:13
alesis-noviksonney2k, overhaul as in the SG* thing?17:14
@sonney2kn_ thingy is just awful wherever it occurs17:14
@sonney2kalesis-novik, with overhaul - yes17:15
heikosonney2k, I have to go now, but compuer is ready, I will impement all the stuff we talked about and send a patch tommorrow ok?17:15
@sonney2kheiko, yes thanks!17:16
heikoalright, then see you!17:16
-!- heiko [865b7525@gateway/web/freenode/ip.] has quit [Quit: Page closed]17:16
-!- f-x [~user@] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]17:19
CIA-32shogun: Soeren Sonnenburg master * r267bed1 / (4 files in 2 dirs): Merge branch 'streaming' of git:// -
-!- blackburn [~blackburn@] has joined #shogun17:52
blackburnmy keyoard =(17:59
-!- alesis-novik [~alesis@] has quit [Quit: biab]18:18
-!- f-x [~user@] has joined #shogun19:19
@sonney2kblackburn, it works!20:04
blackburnwhat works?20:05
@sonney2kI will now valgrind / check the static interfaces20:06
blackburnI have ()roked my key()oard :D20:07
blackburnand now I cant type a letter20:08
blackburnhmm I will type it like 620:08
blackburn6aseclass lol20:13
-!- f-x [~user@] has quit [Read error: Connection reset by peer]22:14
-!- f-x [~user@] has joined #shogun22:15
@sonney2kblackburn, around?23:17
blackburnsonney2k: yes23:17
@sonney2kblackburn, I think I now know why java_modular doesn't compile23:17
@sonney2kproblem is that we have DotFeatures and there a SGVector<float64_t> get_feature_vector() is defined23:18
@sonney2k(not virtual so local to that class)23:18
blackburnin simple features too?23:18
@sonney2know we have the same function in simplefeautres but with SGVector<ST> as return arg23:19
@sonney2k-> clash23:19
@sonney2kin java for some reason23:19
blackburnI see23:19
blackburndo we really need it in DotFeatures?23:19
@sonney2kblackburn, the issue is that some dotfeatures don't provide a get_feature_vector23:19
@sonney2kand indeed it is not needed as one can just call add_to_dense_vec and that will give one the feature vector23:20
blackburnmay 6e just do it virtual?23:20
@sonney2kblackburn, this won't work for sure23:21
blackburndamn it is pretty 6ad to type without 623:21
blackburnstupid me23:21
@sonney2kI mean then we definitely have e.g. SGVector<float64_t> in ther and SGVector<bool> get_feature_vector(int)23:21
blackburnah I see23:21
@sonney2kso we need a rename23:21
blackburnSGVector<float64_t> get_feature_vector()23:22
blackburnwhat it returns?23:22
@sonney2khow about compute_feature_vector?23:22
blackburnwith int?23:22
@sonney2kI mean you give it the vector number23:22
@sonney2kand it returns the vector23:22
blackburnyes I understood23:22
@sonney2kblackburn, such problem could not occur when we had functions like void get_feature_vector(some_type_here)23:23
@sonney2kbut returning different types is a problem23:23
@sonney2kso what do you think? rename to compute_*23:24
blackburnwell just do it compute_*23:24
@sonney2kthe only problem is that it is not so obvious that this will return something23:25
@sonney2kget_* is a better name for that23:25
@sonney2kbut hey - on th eother side get_* sounds like it is already there23:26
@sonney2kor get_computed_*23:26
blackburnget_computed - like it23:26
* sonney2k is pondering23:27
@sonney2kI guess I will have to change 1 million classes23:27
@sonney2k*_my_ass_* :)23:27
blackburnwhy? is it used widely?23:27
@sonney2kblackburn, for some reason some people use this (including yourself :)23:28
blackburnsonney2k: I just wondering where it used23:28
blackburnin SVMs?23:28
blackburnI removed get_feature_vector from gaussiannaive6ayes23:29
blackburnusing feature_matrix now23:29
@sonney2kbut seriously I would call it get_computed_dot_feature_vector and get_computed_dot_feature_matrix23:29
blackburn6ayes haha23:30
-!- blackburn [~blackburn@] has left #shogun []23:30
-!- blackburn [~blackburn@] has joined #shogun23:30
@sonney2kblackburn, yes much better (well use get_feature_vector or get_feature_matrix -> but be aware that sometimes a feature matrix does not exist)23:30
blackburnsonney2k: I hope no6ody will use it for such features23:31
@sonney2kblackburn, I made the name disgusting enough now23:32
@sonney2kso I dont' think anyone will23:32
blackburnokay :)23:33
blackburnhmm what to do next..23:33
@sonney2kgaussiannaivebayes uses it23:33
blackburnsonney2k: uses what?23:33
@sonney2kthat's it23:35
@sonney2kcompiling for all the 100 languages now23:35
@sonney2kdid I say we need a build bot?23:36
@sonney2ksome machine with 64 cores or so23:36
blackburnis it possile?23:36
@sonney2ktoo bad that we don't get 5 per student as organization23:36
@sonney2kthen we could buy sth like that :)23:37
@sonney2kjava built23:37
blackburnsonney2k: will you receive 2.5K for us?23:37
@sonney2kthe university will23:37
@sonney2knot me23:38
blackburnI mean you as organization23:38
@sonney2kI will receive gray hair :D23:38
blackburnsonney2k: which one university?23:39
@sonney2ktu berlin23:39
blackburnis it highly related to tu 6erlin?23:40
blackburnsonney2k: 6tw do you do some teaching activities?23:40
@sonney2kwell our machine learnign department23:41
blackburn*I guess I will receive new key6oard in 1-2 weeks*23:41
@sonney2kso yes23:41
@sonney2kteaching yes23:42
blackburnwhat is do you teach? :)23:42
@sonney2ksome ML :)23:43
blackburnI see ;)23:43
@sonney2klarge scale learning, optimization, bioinformatics, clustering, kernels ...23:43
@sonney2ksuch stuff23:43
blackburnyou could encourage your students to improve our little shogunny23:43
@sonney2knot my students though23:44
@sonney2kthe ones of our prof ...23:44
blackburnwell anyway you can suggest somethig :)23:44
blackburnif you won't have time just suggest them to ask drunk russian for some tasks23:45
blackburnI'm lazy a6out writing 6alltree and spectral clustering23:45
blackburnmay 6e some other student will :D23:46
CIA-32shogun: Soeren Sonnenburg master * r65ac9bf / (13 files in 8 dirs): rename dot features' get_feature_vector/matrix functions to avoid naming clash -
--- Log closed Sat Jun 25 00:00:18 2011

Generated by 2.10.0 by Marius Gedminas - find it at!