--- Log opened Mon Oct 10 00:00:28 2011 | ||
-!- blackburn [~blackburn@85.114.187.69] has quit [Read error: Connection reset by peer] | 00:19 | |
-!- blackburn [~blackburn@85.114.187.69] has joined #shogun | 00:19 | |
-!- blackburn [~blackburn@85.114.187.69] has quit [Quit: Leaving.] | 00:24 | |
-!- av3ngr [96cbdf9c@gateway/web/freenode/ip.150.203.223.156] has quit [Ping timeout: 252 seconds] | 04:30 | |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has joined #shogun | 17:54 | |
-!- blackburn [~blackburn@188.168.4.72] has joined #shogun | 20:28 | |
-!- blackburn [~blackburn@188.168.4.72] has quit [Ping timeout: 258 seconds] | 20:33 | |
-!- blackburn [~blackburn@188.168.4.100] has joined #shogun | 20:34 | |
blackburn | sonney2k: what's up? | 21:36 |
---|---|---|
@sonney2k | making progress on the debian package... | 21:36 |
blackburn | oh that's pretty nice | 21:36 |
blackburn | would you mind making .exe installation? | 21:36 |
blackburn | well windows is a piece of shit but | 21:37 |
@sonney2k | impossilbe | 21:37 |
blackburn | почему? | 21:38 |
blackburn | sonney2k: okay, rpm? | 21:39 |
@sonney2k | it will be some cygwin stuff | 21:39 |
blackburn | sonney2k: I have one problem still to be solved - I can't use dimensionality reduction on say SparseFeatures although it is possible to do | 21:46 |
blackburn | it makes me mad | 21:47 |
blackburn | I can't sleep | 21:47 |
blackburn | I eat wood | 21:47 |
blackburn | ok joke | 21:47 |
@sonney2k | well that's life | 21:53 |
@sonney2k | you cannot have it all | 21:53 |
blackburn | sonney2k: I see one solution but I guess you won't like it | 21:55 |
blackburn | sonney2k: DimensionReductionPreprocessor as proxy of DimensionReductionMachine, but not derived from CMachine | 21:57 |
blackburn | with methods apply(CFeatures*) returning CFeatures* | 21:57 |
blackburn | how it is for you>? | 21:59 |
blackburn | moreover I could even return SimpleFeatures all the time | 21:59 |
blackburn | imagine how useful it would be to input StringFeatures and get SimpleFeatures representing it | 22:00 |
blackburn | ain't it? | 22:00 |
@sonney2k | yes but your solution is not nice :( | 22:04 |
@sonney2k | I think we need the concept of 'converters' for that | 22:05 |
@sonney2k | that take as input feature type X and are allowed to return Y | 22:05 |
blackburn | think about it please, I can't see any better solution | 22:06 |
blackburn | sonney2k: I'm now using svn at my job heh | 22:13 |
blackburn | ugly thing, don't know how you had used it before gsoc | 22:13 |
@sonney2k | blackburn, why not have a converter class that has an apply method and gets features in and returns features | 22:15 |
blackburn | sonney2k: it is the same I propose | 22:16 |
@sonney2k | no much better - more general and more clear | 22:16 |
blackburn | how could we call it? | 22:16 |
@sonney2k | Converter | 22:17 |
blackburn | hah | 22:17 |
blackburn | so, DimensionReductionConverter? | 22:17 |
blackburn | I like it | 22:18 |
@sonney2k | what I don't understan though is how you proxy w/ preprocs | 22:18 |
blackburn | easy | 22:18 |
blackburn | DimensionReductionPreprocessor(LLE) | 22:18 |
blackburn | or | 22:18 |
blackburn | preproc.set_technique(LLE) | 22:19 |
blackburn | preproc.set_technique(BLABLAE) | 22:19 |
blackburn | is it bad for you? | 22:19 |
blackburn | it just delegate responsibilities to converter | 22:20 |
blackburn | I have no in-place algorithms | 22:20 |
blackburn | so proxy will just take and replace feature matrix | 22:20 |
blackburn | after converting | 22:20 |
blackburn | what I would say with some confidence - NLDR techniques are used as converters more often than as preprocessors | 22:22 |
blackburn | with proxy we will still able to built even chains of these techniques (if it is needed, it is a little crazy thought) | 22:23 |
@sonney2k | blackburn, couldn't you add an inplace modify option? | 22:24 |
@sonney2k | It certainly is not supported to dealloc / alloc the features in the preproc | 22:25 |
blackburn | sonney2k: in most cases it is impossible | 22:25 |
blackburn | I do it in every algorithm | 22:25 |
blackburn | I change feature matrix | 22:25 |
blackburn | sonney2k: very difficult challenge to replace it in-place | 22:27 |
blackburn | and well | 22:27 |
blackburn | it is not that bad to have 2 matrices of N*dim while I have 1-2 matrices of N*N, right? | 22:28 |
blackburn | sonney2k: or do you mean any other cons of replacing feature matrices? | 22:29 |
@sonney2k | blackburn, well why not create a tmp copy then and do the mdoification later inplace | 22:29 |
@sonney2k | (if there is no easy other way) | 22:30 |
@sonney2k | certainly ok for most of the applications | 22:30 |
blackburn | what do you mean? | 22:30 |
blackburn | tmp copy of ? | 22:30 |
@sonney2k | other preprocessors didnt' need any helper array | 22:30 |
@sonney2k | dimred feature matrix | 22:30 |
blackburn | I need a LOT of arrays | 22:30 |
blackburn | and almost 2x memory for 2 threads | 22:31 |
blackburn | etc | 22:31 |
blackburn | so what do you suggest? sometimes it is a little more possible but I would have to make transpose of non-square matrix in-place and etc | 22:32 |
blackburn | I have no time now for such crazy algos | 22:32 |
blackburn | my algos are not that easy :) | 22:34 |
blackburn | you know I guess | 22:34 |
blackburn | *valgrind became my favourite tool* | 22:40 |
@sonney2k | blackburn, well when you have a copy you can just copy it over to the original matrix | 22:42 |
@sonney2k | so no problem | 22:42 |
blackburn | it is just technical issues | 22:43 |
blackburn | not to use set_feature_matrix but overwrite gotten feature matrix pointer | 22:43 |
blackburn | sonney2k: got an idea | 22:44 |
blackburn | what if check sizes at set_feature_matrix | 22:44 |
blackburn | why to delete available feature matrix? | 22:45 |
blackburn | if it have the same sizes | 22:45 |
@sonney2k | blackburn, others might use the ptr | 22:45 |
@sonney2k | so I don't like the idea of deleting it... | 22:46 |
@sonney2k | better adjust shape (shrink) | 22:46 |
@sonney2k | and waste a littley memory | 22:46 |
blackburn | sonney2k: currently in set_feature_matrix we free it | 22:46 |
blackburn | why not to realloc it if necessary | 22:46 |
blackburn | or just copy content if have the same size | 22:46 |
blackburn | ? | 22:47 |
blackburn | it will make my preprocessors work in-place | 22:47 |
blackburn | at least | 22:47 |
blackburn | not in-place but with realloc | 22:47 |
blackburn | I see no reason not to do it | 22:51 |
@sonney2k | blackburn, one reason would be someone having a ptr to e.g. a SimpleFeature object | 22:53 |
blackburn | and? | 22:53 |
@sonney2k | then changing the data ptr might make things crash | 22:53 |
blackburn | sonney2k: it is being changed now!!1 | 22:54 |
@sonney2k | blackburn, you do this in your prepro*s? | 22:54 |
blackburn | I just call set_feature_matrix | 22:54 |
blackburn | of simplefeatures | 22:54 |
blackburn | just take a look on SimpleFeatures.h:273 | 22:55 |
blackburn | I suggest not to free it but realloc! | 22:55 |
@sonney2k | ohh well we can give it a try... | 22:55 |
blackburn | and if it was 400*3 and now should be 200*6 | 22:56 |
blackburn | we even don't need to realloc it | 22:56 |
CIA-3 | shogun: Sergey Lisitsyn * r00b7950 / (19 files): Cleaned DR preprocessors, removed effective target dim computation - http://git.io/QsWUaw | 23:12 |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has quit [Ping timeout: 276 seconds] | 23:13 | |
blackburn | nono | 23:14 |
blackburn | I don't know what is better :( | 23:15 |
blackburn | copying is slower than freeing | 23:15 |
CIA-3 | shogun: Soeren Sonnenburg * r77efd4c / src/interfaces/elwms_static/elwms.cpp : Fix include paths for elwms interface. - http://git.io/W2YdOQ | 23:18 |
blackburn | sonney2k: it's an achievement there is no memory leaks with my preprocs :) | 23:20 |
blackburn | btw I got an idea of memory check test | 23:20 |
blackburn | we should run every test with valgrind --leak-check=full and search for 'shogun' | 23:21 |
blackburn | sonney2k: TU decided not to give you buildbots? | 23:21 |
@sonney2k | blackburn, it takes time | 23:21 |
blackburn | btw I don't understand why my code have copyright of TU | 23:22 |
blackburn | :D | 23:22 |
blackburn | sonney2k: why is it a TU property? | 23:22 |
@sonney2k | blackburn, all the new code you write (outside GSoC) can have some other code | 23:22 |
@sonney2k | license | 23:22 |
blackburn | GSoC code -> TU copyright? | 23:23 |
blackburn | I'm asking because I can't see TU care any about us | 23:23 |
@sonney2k | argh, I still have to update all copyright stuff for the debian package | 23:24 |
@sonney2k | blackburn, patience young padawan | 23:24 |
blackburn | I'm not about buildbots only | 23:24 |
blackburn | why GSoC code is a TU property while TU doesn't care about it anyhow? :) | 23:24 |
@sonney2k | it is MPI / TU | 23:25 |
blackburn | what is MPI/TU? | 23:26 |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 23:27 | |
blackburn | I'm pretty interested if I can't change it some day | 23:27 |
blackburn | I can't, right? | 23:27 |
blackburn | sonney2k: why my code is licensed over orgs I'm not a member of? :) | 23:28 |
@sonney2k | blackburn, indeed you cannot | 23:29 |
@sonney2k | because you took part in gsoc and agreed to the gsoc terms | 23:29 |
@sonney2k | all new code you wrote after gsoc is 100% yours though :) | 23:30 |
blackburn | sonney2k: impossible to determine what I have wrote after gsoc | 23:30 |
@sonney2k | git log | 23:30 |
blackburn | wrote = new file? | 23:30 |
blackburn | then I could remove that shit from KLLE and KLTSA | 23:31 |
blackburn | ah nevermind but that's strange | 23:32 |
blackburn | you do all the work and receive no support neither from MPI nor TU | 23:32 |
@sonney2k | well we receive support - not as much as you would like but we do | 23:34 |
blackburn | what is the support | 23:34 |
blackburn | ? | 23:34 |
blackburn | okay I have removed copyrights from new algos ;P | 23:36 |
CIA-3 | shogun: Sergey Lisitsyn * rf6d08c2 / (4 files): Removed wrong copyrights from preprocessors developed after GSoC - http://git.io/s9V8TA | 23:36 |
blackburn | fuck I did it in gsoc | 23:38 |
@sonney2k | blackburn, you should put Copyright (C) 2011 then instead of the writting stuff | 23:38 |
blackburn | revert | 23:38 |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Ping timeout: 258 seconds] | 23:38 | |
blackburn | yeah that's odd I'll better sleep | 23:39 |
CIA-3 | shogun: Sergey Lisitsyn * ra92e6b4 / (4 files): | 23:39 |
CIA-3 | shogun: Revert "Removed wrong copyrights from preprocessors developed after GSoC" | 23:39 |
CIA-3 | shogun: This reverts commit f6d08c2226fd1b297744a4c7379fc7f180d6d60e. - http://git.io/YnvpnQ | 23:39 |
@sonney2k | ok | 23:39 |
blackburn | I still see no link shogun <-> TU | 23:40 |
blackburn | except of you was there | 23:40 |
@sonney2k | blackburn, we will get 3 buildbots just be patient | 23:40 |
blackburn | hah | 23:41 |
blackburn | it would be better to suggest students to take participance | 23:42 |
blackburn | I will never forget methods I have implemented | 23:42 |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 23:47 | |
--- Log closed Tue Oct 11 00:00:32 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!