| --- Log opened Thu May 24 00:00:41 2012 | ||
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 02:03 | |
| -!- wiking [~wiking@78-23-189-112.access.telenet.be] has joined #shogun | 02:08 | |
| -!- wiking [~wiking@78-23-189-112.access.telenet.be] has quit [Changing host] | 02:08 | |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 02:08 | |
| -!- av3ngr [av3ngr@nat/redhat/x-dnxruybggcigxdmo] has joined #shogun | 02:16 | |
| -!- av3ngr [av3ngr@nat/redhat/x-dnxruybggcigxdmo] has quit [Client Quit] | 02:16 | |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 03:04 | |
| -!- vikram360 [~vikram360@117.192.161.189] has quit [Ping timeout: 256 seconds] | 07:05 | |
| -!- vikram360 [~vikram360@117.192.175.147] has joined #shogun | 07:05 | |
| -!- pluskid [~pluskid@li400-235.members.linode.com] has joined #shogun | 07:37 | |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 08:11 | |
| CIA-113 | shogun: Soeren Sonnenburg master * r95c124d / (24 files): fix most of octave_modular examples for new labels - http://git.io/cbIPiw | 08:20 |
|---|---|---|
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 08:33 | |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 08:34 | |
| -!- uricamic [~uricamic@2001:718:2:1634:ac88:5a8e:28de:2e53] has joined #shogun | 08:53 | |
| -!- wiking [~wiking@78-23-189-112.access.telenet.be] has joined #shogun | 08:54 | |
| -!- wiking [~wiking@78-23-189-112.access.telenet.be] has quit [Changing host] | 08:54 | |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 08:54 | |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 265 seconds] | 09:03 | |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 09:16 | |
| -!- wiking [~wiking@78-23-189-112.access.telenet.be] has joined #shogun | 09:23 | |
| -!- wiking [~wiking@78-23-189-112.access.telenet.be] has quit [Changing host] | 09:23 | |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 09:23 | |
| -!- blackburn [~blackburn@188.122.250.167] has joined #shogun | 10:54 | |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 11:39 | |
| -!- vikram360 [~vikram360@117.192.175.147] has quit [Ping timeout: 245 seconds] | 11:50 | |
| -!- n4nd0 [~nando@n155-p45.kthopen.kth.se] has joined #shogun | 11:57 | |
| -!- wiking [~wiking@we02c096.ugent.be] has joined #shogun | 11:58 | |
| -!- wiking [~wiking@we02c096.ugent.be] has quit [Changing host] | 11:58 | |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 11:58 | |
| n4nd0 | pluskid: hey! did you find the version of the paper you talked about yesterday? | 11:58 |
| n4nd0 | pluskid: I am in the university now so I can try to look for it otherwise | 11:58 |
| pluskid | n4nd0: no, please try | 11:59 |
| pluskid | I don't have easy access to the school network after graduation... :( | 11:59 |
| n4nd0 | pluskid: all right, got it :) | 12:00 |
| n4nd0 | pluskid: do you want me to mail it to you, dropbox, or? | 12:00 |
| pluskid | n4nd0: mail me, thanks! | 12:02 |
| pluskid | :D | 12:02 |
| n4nd0 | all right, it should be in your mailbox soon | 12:02 |
| n4nd0 | pluskid: let me know if it is the correct one | 12:02 |
| pluskid | cool! | 12:02 |
| pluskid | n4nd0: awesome! It's the one I'm looking for | 12:03 |
| n4nd0 | pluskid: nice, and I forgot to say, you are welcome ;) | 12:03 |
| -!- imeanwow [46e7e626@gateway/web/freenode/ip.70.231.230.38] has joined #shogun | 12:03 | |
| pluskid | :D | 12:03 |
| imeanwow | lol | 12:03 |
| imeanwow | I accidently typed #shogun when I tried to name myself Shogun | 12:04 |
| imeanwow | and that mistake brought me here. | 12:04 |
| pluskid | welcome | 12:04 |
| -!- pluskid is now known as Shogun | 12:04 | |
| imeanwow | lol | 12:05 |
| blackburn | n4nd0: Shogun: what is the paper? | 12:09 |
| n4nd0 | blackburn: sensitive ECOC | 12:09 |
| n4nd0 | blackburn: http://www.springerlink.com/content/n30xeen83d81ylbu/?MUD=MP | 12:09 |
| blackburn | some data-dependant ecoc scheme? | 12:09 |
| blackburn | hmm can you mail to me too? :) | 12:10 |
| n4nd0 | sure | 12:10 |
| Shogun | blackburn: don't know, haven't read yet :p | 12:11 |
| -!- Shogun is now known as pluskid | 12:11 | |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 245 seconds] | 12:18 | |
| blackburn | n4nd0: when do you think you will be able to merge structured labels thing? | 12:20 |
| n4nd0 | blackburn: now it is in compilable state but there are things that probably need to be completed | 12:21 |
| n4nd0 | blackburn: what state do you want it to be in order to be merged? | 12:21 |
| blackburn | hmm no idea - what is uncompleted in SO labels? | 12:22 |
| n4nd0 | I just have a basic interface now | 12:23 |
| n4nd0 | once I start using it for specific examples, functionality needs will pop up | 12:23 |
| n4nd0 | blackburn: for me, the sooner we merge it, the better. It feels safer | 12:25 |
| blackburn | yes - but can't you provide all the needed functionality in that interface? | 12:25 |
| n4nd0 | blackburn: it could be a good idea to wait since I may not have a perfect idea of all we need there a priori | 12:26 |
| blackburn | that's why you have mentor :) | 12:29 |
| blackburn | however yes hard to predict everything | 12:29 |
| n4nd0 | what has it happened in github? the comments in the interface files seem to be wrong indented | 12:30 |
| blackburn | where? | 12:33 |
| n4nd0 | https://github.com/shogun-toolbox/shogun/blob/master/src/shogun/classifier/LDA.h | 12:34 |
| n4nd0 | I have opened some of them at random and they look so | 12:35 |
| pluskid | looks fine here | 12:42 |
| n4nd0 | pluskid: mmm really? it appears in the browser, it should be the same for all I guess | 12:43 |
| pluskid | all comments are of the same indent in my browser | 12:44 |
| imeanwow | What is right or what is wrong? When you realize that you possess a highly powerful set of skills | 12:44 |
| -!- wiking [~wiking@huwico/staff/wiking] has joined #shogun | 12:54 | |
| -!- blackburn [~blackburn@188.122.250.167] has quit [Ping timeout: 248 seconds] | 13:16 | |
| sonne|work | n4nd0: my take on this is that I would rather merge early and often as long as your code compiles and doesn't break anything | 14:00 |
| sonne|work | and as you have seen - nothing in shogun is set in stone... | 14:00 |
| sonne|work | but some things are harder to correct than others | 14:00 |
| -!- vikram360 [~vikram360@117.192.181.10] has joined #shogun | 14:01 | |
| n4nd0 | sonne|work: it compiles well and I have nothed nothing breaking here :) | 14:01 |
| sonne|work | then pull request and I merge | 14:01 |
| n4nd0 | sonne|work: ok, I will rebase first to introduce changes from yesterday midday | 14:02 |
| n4nd0 | sonne|work: I think it is a good idea to merge since in that way uricamic, wiking and I know how each other is working | 14:05 |
| n4nd0 | what will be good since our projects are somewhat related | 14:05 |
| wiking | anybody seen alex around today? | 14:07 |
| sonne|work | wiking: no ... | 14:08 |
| wiking | :( | 14:08 |
| wiking | we suppose to have a meeting now | 14:08 |
| wiking | or at least today | 14:08 |
| sonne|work | wiking: hmmmh? | 14:08 |
| sonne|work | email him :) | 14:08 |
| wiking | sonne|work: yeah we've agreed it via email ;P | 14:09 |
| sonne|work | no email him now again reminding him | 14:09 |
| n4nd0 | by the way, does any of you see this warning when compiling? | 14:12 |
| n4nd0 | ./../shogun/lib/DynamicArray.h:296: Warning 516: Overloaded method shogun::CDynamicArray< char >::element(int32_t,int32_t,int32_t) ignored, | 14:12 |
| n4nd0 | ../../shogun/lib/DynamicArray.h:284: Warning 516: using shogun::CDynamicArray< char >::element(int32_t,int32_t,int32_t) const instead. | 14:12 |
| n4nd0 | it appears quite a lot of times here | 14:12 |
| sonne|work | n4nd0: with swig you mean? | 14:35 |
| sonne|work | yeah no idea why we need the const one... | 14:35 |
| n4nd0 | sonne|work: I don't know if it is a problem with SWIG or something different | 14:35 |
| n4nd0 | and I am not sure if it is something that happens to me locally | 14:36 |
| sonne|work | n4nd0: well if you compile with ./configure --interface= | 14:36 |
| sonne|work | this warning shouldn't be there | 14:36 |
| n4nd0 | I am compiling with ./configure --interface=python_modular --disable-optimization --disable-doxygen | 14:36 |
| n4nd0 | mmmm maybe --integerface=pyton_modular was not there the last time | 14:37 |
| n4nd0 | sonne|work: ok, I've just compiled with --interfaces= and no warnings :) thank you! | 14:43 |
| sonne|work | n4nd0: well you usually (before doing a pull request) should compile it with one static and one modular interface | 14:44 |
| sonne|work | otherwise lots of errors slip through | 14:44 |
| n4nd0 | ok | 14:45 |
| sonne|work | n4nd0: can we have a better name than VanillaStructuredOutputMachine? | 14:48 |
| -!- wiking [~wiking@huwico/staff/wiking] has quit [Quit: wiking] | 14:48 | |
| sonne|work | what is a vanilla machine? | 14:48 |
| sonne|work | ice cream :) | 14:49 |
| n4nd0 | haha | 14:49 |
| n4nd0 | we are using there the optimization algorithm that we have designated as "vanilla" | 14:49 |
| n4nd0 | I guess because it is the most straighforward or naive one, not because of its taste :P | 14:49 |
| n4nd0 | mmm let me see | 14:50 |
| sonne|work | n4nd0: I first thought you meant the SOMachine | 14:51 |
| sonne|work | but then I realized that there is SOMachine -> linearsomachien -> vanillasomachine | 14:51 |
| n4nd0 | yes | 14:51 |
| n4nd0 | sonne|work: check here http://iglesiashogun.wordpress.com/ | 14:52 |
| n4nd0 | there is a class diagram in the last post that will make you see the design faster | 14:52 |
| n4nd0 | sonne|work: now I realize that it probably makes sense to have the weight vector in LinearSOMachine instead of in VanillaSOMachine | 14:54 |
| sonne|work | yeah ok but we need some better term than vanilla | 14:57 |
| sonne|work | n4nd0: anyway I am merging right now but please fix both the 'vanilla' and w -> linearsomachine | 14:58 |
| n4nd0 | sonne|work: any suggestion instead of vanilla? | 14:58 |
| CIA-113 | shogun: Soeren Sonnenburg master * r8b0279c / (28 files in 5 dirs): Merge pull request #551 from iglesias/so (+18 more commits...) - http://git.io/rBhtYw | 14:59 |
| sonne|work | n4nd0: what does 'vanilla' do ? | 14:59 |
| sonne|work | n4nd0: nice blog | 15:00 |
| sonne|work | I have to add it to planet tonight | 15:00 |
| n4nd0 | sonne|work: it is the algorithm Nico wrote in the pdf he sent around | 15:01 |
| sonne|work | n4nd0: which one | 15:01 |
| sonne|work | there were many :) | 15:02 |
| n4nd0 | sonne|work: http://dl.dropbox.com/u/11020840/shogun/so.pdf | 15:02 |
| n4nd0 | in the second page of the document | 15:03 |
| n4nd0 | Algorithm 1 (the only algorithm there) | 15:03 |
| sonne|work | alright the primal solver | 15:03 |
| sonne|work | unconstrained optimzation problem ? | 15:04 |
| n4nd0 | there are constraints | 15:05 |
| n4nd0 | although they are not written in the algorithm | 15:05 |
| sonne|work | ok | 15:05 |
| n4nd0 | they appear in equation (1), using the output of init_op | 15:05 |
| n4nd0 | but I think that, even if the formulation that appears there is for the primal, this same algorithm can be used for the dual | 15:06 |
| sonne|work | n4nd0: I think you should change then name completely then | 15:06 |
| sonne|work | I mean drop the SOMachine part | 15:06 |
| n4nd0 | in the paper they do it with the dual IIRC | 15:06 |
| sonne|work | and invent something like libsvm :) | 15:07 |
| n4nd0 | I would like to have the SO part though | 15:08 |
| sonne|work | n4nd0: will you be using mosek for it? | 15:08 |
| n4nd0 | this is something that works for SO since we are using this Delta loss function and the combined featuers | 15:08 |
| n4nd0 | sonne|work: hopefully | 15:08 |
| n4nd0 | I need some help with the formulation of the optimization problem, I am supposed to talk to Nico about it now around here ... | 15:09 |
| sonne|work | n4nd0: yeah but suffixing with the machine name unecessary makes the name longer | 15:09 |
| -!- karlnapf [~heiko@host86-176-125-255.range86-176.btcentralplus.com] has joined #shogun | 15:09 | |
| sonne|work | I mean we could have LiblinearLinearClassifierMachine otherwise | 15:10 |
| n4nd0 | sonne|work: haha ok you convinced me | 15:10 |
| -!- karlnapf1 [~heiko@host86-177-183-14.range86-177.btcentralplus.com] has joined #shogun | 15:12 | |
| -!- karlnapf [~heiko@host86-176-125-255.range86-176.btcentralplus.com] has quit [Ping timeout: 260 seconds] | 15:14 | |
| -!- n4nd0 [~nando@n155-p45.kthopen.kth.se] has quit [Ping timeout: 244 seconds] | 15:31 | |
| -!- n4nd0 [~nando@n155-p45.kthopen.kth.se] has joined #shogun | 15:41 | |
| -!- nicococo [~nico@lacedcoffee.ml.tu-berlin.de] has joined #shogun | 15:58 | |
| nicococo | hi n4nd0 | 16:00 |
| n4nd0 | hey nicococo, how is it going? | 16:01 |
| nicococo | medium to well :) | 16:01 |
| nicococo | how is the toolbox progressing? | 16:01 |
| n4nd0 | so so+ | 16:02 |
| n4nd0 | I am kind of stuck with the optimization part | 16:02 |
| nicococo | hehe.. what can i do to help speed it up? | 16:02 |
| n4nd0 | so I don't know how to translate a couple of terms in our optimization problem to mosek | 16:03 |
| n4nd0 | the first one I already asked you by mail, but I think I didn't get that much the answer | 16:03 |
| n4nd0 | so we have this | 16:03 |
| n4nd0 | min_w 1/2 w'?C?w + sum_i l(...) | 16:04 |
| nicococo | jupp | 16:04 |
| n4nd0 | where l is loss function evaluated in this expression max_y | 16:04 |
| n4nd0 | you know which one I mean right? | 16:04 |
| nicococo | okay and the optimizer has the following form: | 16:04 |
| nicococo | min_x 0.5 x'Hx + f'x .. right | 16:05 |
| nicococo | ? | 16:05 |
| n4nd0 | yes | 16:05 |
| n4nd0 | with a constant term too | 16:05 |
| n4nd0 | but yes, that's it | 16:05 |
| nicococo | the trick here is: 1. we don't use the unconstraint version | 16:05 |
| nicococo | which means we have: 0.5w'Cw + sum_i slack_i | 16:05 |
| n4nd0 | ok | 16:06 |
| nicococo | all you have to do know is to add one constraint per iteration per example to the Ax <= b | 16:06 |
| nicococo | now, we have to setup the optimization variable x which is a block variable | 16:07 |
| nicococo | x = [w, slacks] | 16:07 |
| nicococo | then f = ones(N,1) | 16:08 |
| sonne|work | nicococo: hey there - do you have an idea for a short but more reasonable name than VanillaSOMachine? | 16:08 |
| nicococo | i like vanilla... don't you | 16:08 |
| sonne|work | no the name is too long | 16:08 |
| n4nd0 | VanillaSVM | 16:09 |
| sonne|work | we have CPLEXSvm :) | 16:09 |
| n4nd0 | or VanillaSOSVM | 16:09 |
| nicococo | fine with me | 16:09 |
| sonne|work | not vanilla though | 16:09 |
| sonne|work | it doesn't say anything | 16:09 |
| n4nd0 | it says that it is basic | 16:10 |
| sonne|work | nicococo: don't you have a name for that problem? | 16:10 |
| nicococo | BasicPrimalSOSVM | 16:10 |
| nicococo | ? | 16:10 |
| sonne|work | a lot better to digest | 16:10 |
| sonne|work | does it need SVM? | 16:10 |
| sonne|work | in the name? | 16:10 |
| sonne|work | otherwise we take this | 16:11 |
| nicococo | BasicPrimal is not a good name without SOSVM, right? | 16:11 |
| sonne|work | err if yes | 16:11 |
| sonne|work | nicococo: you can also think up names like svmlight, torch, libsvm or whatever | 16:12 |
| n4nd0 | nicococo: I don't understand why f turns out to be a vector of ones | 16:12 |
| nicococo | cool, what abaout VanillaSOSVM ;) | 16:12 |
| n4nd0 | :D | 16:12 |
| sonne|work | nicococo: when we meet again next time ....DANGER | 16:13 |
| sonne|work | nicococo: will this use mosek? | 16:13 |
| nicococo | i'm not really scared :) and yes, mosek it is | 16:13 |
| sonne|work | PrimalMosekSOSVM then | 16:13 |
| nicococo | alright | 16:14 |
| n4nd0 | agree | 16:14 |
| nicococo | okay back to ones vector? | 16:15 |
| n4nd0 | nicococo: so how it turns out that f = ones(N,1)? | 16:15 |
| n4nd0 | for me that looks like | 16:15 |
| nicococo | youre right it is a zero ones vector | 16:16 |
| n4nd0 | min w'Cw + w | 16:16 |
| sonne|work | nicococo: did you see Alexander today? wiking was looking for him said that they wanted to meet in chat.. | 16:16 |
| nicococo | no, we have x = [w;slacks] | 16:16 |
| nicococo | sonney, i can give him a hint | 16:16 |
| n4nd0 | nicococo: zero ones vector? | 16:16 |
| nicococo | okay, so x = [w;slacks] and min_x 0.5 x'Hx + f'x | 16:17 |
| n4nd0 | slacks is empty at the beginning or? | 16:17 |
| nicococo | if f =[zeros(len(w),1); ones(N,1)] then we have sum_i slacks_i | 16:17 |
| n4nd0 | ok | 16:18 |
| n4nd0 | I understand what you mean :) | 16:18 |
| nicococo | cool ;) | 16:18 |
| n4nd0 | so this vector x has always the same dimensions | 16:19 |
| nicococo | then H = [C,zeros(N);zeros(N),zeros(N)] right? | 16:19 |
| nicococo | x has always dim = len(w) + num_examples | 16:19 |
| n4nd0 | agree | 16:19 |
| nicococo | when you generate a new constraint then you add a line to A and b | 16:20 |
| n4nd0 | mmm, ok wait a moment there | 16:22 |
| n4nd0 | we have two types of constraints | 16:22 |
| nicococo | where A = [-psi_i +psi_i(y_argmax) -delta_i(y_argmax), ...] | 16:22 |
| nicococo | two types? | 16:22 |
| n4nd0 | Aw <= a | 16:22 |
| n4nd0 | and | 16:22 |
| n4nd0 | Bw = b | 16:22 |
| nicococo | no, we have Ax <= b | 16:23 |
| nicococo | and thats enough we don't need the equality constraint | 16:23 |
| n4nd0 | mmmm | 16:23 |
| nicococo | and we have a lower bound for slacks | 16:23 |
| n4nd0 | yes | 16:23 |
| n4nd0 | so is it like Ax <= b includes also the equality constraint? | 16:24 |
| n4nd0 | or it shouldn't be there in the pdf | 16:24 |
| nicococo | which equality constraint you mean? | 16:24 |
| n4nd0 | in the pdf you wrote | 16:24 |
| n4nd0 | nicococo: http://dl.dropbox.com/u/11020840/shogun/so.pdf | 16:25 |
| nicococo | oh.. in the pdf was a very general version .. all constraints there are for applications only.. | 16:25 |
| nicococo | here: 1. we drop all the constaints i wrote in the pdf | 16:25 |
| nicococo | and 2. we make the unconstraint version constraint again by introducing slacks | 16:25 |
| nicococo | hence, we have inequality constraints again | 16:26 |
| nicococo | is that clear? | 16:26 |
| n4nd0 | I don't understand why to introduce the slacks as in x = [w; slacks] | 16:27 |
| n4nd0 | makes that inequality contraints like Ax <= b appear | 16:27 |
| nicococo | we introduce the slacks because otherwise we cannot solve it with the QP solver | 16:27 |
| n4nd0 | yes, I understand that | 16:27 |
| n4nd0 | but I don't understand why if we have something without contraints | 16:28 |
| n4nd0 | why then Ax <= b appears | 16:28 |
| n4nd0 | I have clear the min x'Hx + f'x part | 16:28 |
| n4nd0 | how we have get there from min w'Cw + sum_i ... | 16:29 |
| nicococo | sometimes in some application you may want to introduce some restrictions to your variables in w.. like the w(20) <= 10 or whatever | 16:29 |
| nicococo | was that your question? | 16:29 |
| n4nd0 | but those are lb <= w <= ub constraints | 16:30 |
| nicococo | ..or how do we get from the unconstraint to the constraint version.. | 16:30 |
| nicococo | (ps. what about w(20)-w(19)<=10) | 16:31 |
| n4nd0 | ok | 16:31 |
| n4nd0 | so what about A and b in Ax <= b | 16:32 |
| n4nd0 | you said something about adding rows and cols | 16:32 |
| nicococo | w'psi_i - w'psi_yhat >= delta_yi_yhat - slack_i | 16:32 |
| nicococo | well, this is one line in Ax <= b, right? | 16:38 |
| n4nd0 | yes | 16:38 |
| nicococo | okay, so is everything clear? | 16:38 |
| n4nd0 | nicococo: yeah ... kind of :) | 16:39 |
| nicococo | if you have more questions, please ask :) | 16:39 |
| n4nd0 | nicococo: I will try to formulate the problem for mosek | 16:40 |
| n4nd0 | I think I should be able to do it with this new info you have given to me | 16:40 |
| n4nd0 | although I don't know where some of the terms given by init_op, like A, a, B and b are used | 16:41 |
| nicococo | the new constraint would be: -delta >= w'psi_y_hat - w'psi_i - slack_i | 16:41 |
| n4nd0 | and this is a bit dangerous now since there are two different things called A (and the same for b) | 16:41 |
| nicococo | ohh.. right. Ax <= a | 16:42 |
| nicococo | in many applications A,a,B,b are empty in the beginning and C=eye(len(w)) | 16:43 |
| n4nd0 | nicococo: ok, I think I can forget about them for the moment | 16:44 |
| nicococo | and in PrimalMosekSOSVM we don't handle equality constraints Bx = b | 16:44 |
| n4nd0 | agree? | 16:44 |
| nicococo | agree | 16:44 |
| n4nd0 | I don't have clear one point | 16:44 |
| n4nd0 | about Ax <= b | 16:44 |
| n4nd0 | what are the training examples that have to be included there? | 16:45 |
| nicococo | psi(x_i,y_i) ? | 16:46 |
| n4nd0 | let's put it in another way | 16:46 |
| n4nd0 | Ax <= b represents constraints that we use like w'psi_y_Hat - w'psi_i <= -delta + slack_i | 16:47 |
| n4nd0 | each i is a training example | 16:47 |
| n4nd0 | what are the training examples that we have to include? | 16:47 |
| n4nd0 | mmm I don't think I am making it very clear :S | 16:47 |
| nicococo | agree | 16:47 |
| n4nd0 | :D | 16:47 |
| nicococo | :) | 16:47 |
| n4nd0 | ok | 16:48 |
| nicococo | aah!!! | 16:48 |
| n4nd0 | A and b in Ax <=b, do they always have the same size? | 16:48 |
| nicococo | i think i know what you mean | 16:48 |
| nicococo | A is growing one line per iteration and b is growing too | 16:48 |
| n4nd0 | ok | 16:48 |
| n4nd0 | in the first iteration, the first time we solve the QP, how are they? | 16:49 |
| nicococo | in the first iteration, you first add num_examples constraints therefor A has num_examples lines | 16:49 |
| nicococo | then you solve it | 16:50 |
| n4nd0 | how does A and b look like at the beginning? | 16:50 |
| n4nd0 | if we have no w initially, how do we get y_hat? | 16:50 |
| nicococo | w =zeros | 16:51 |
| nicococo | or rand or something else we get from the pre-solver | 16:52 |
| n4nd0 | ok, and we just predict with that | 16:52 |
| n4nd0 | ok | 16:52 |
| nicococo | yepp, the first iteration will give us really bad estimates certainly | 16:52 |
| n4nd0 | and then, what are the constraints that are added, one per iteration? | 16:53 |
| nicococo | one per example per iteration | 16:53 |
| n4nd0 | which example is added? the one with max loss? | 16:54 |
| nicococo | for each example add constraint if it produces more slack than all the constraints added for the current example | 16:55 |
| n4nd0 | so in the pdf, Vanilla primal-SO-SVM | 16:56 |
| n4nd0 | this is what is done in lines 7-9 | 16:57 |
| n4nd0 | ? | 16:57 |
| nicococo | yes, looks cryptic but is actually easy | 16:58 |
| -!- uricamic [~uricamic@2001:718:2:1634:ac88:5a8e:28de:2e53] has quit [Quit: Leaving.] | 16:58 | |
| -!- pluskid [~pluskid@li400-235.members.linode.com] has quit [Quit: Leaving] | 16:58 | |
| nicococo | (there might be some signs that are wrong ;) ) | 16:59 |
| n4nd0 | nicococo: ok, let me try to give it a go then | 17:00 |
| nicococo | and a w'psi_i term is missing :) | 17:00 |
| n4nd0 | nicococo: and I will get back to you so we can correct mistakes ;) | 17:00 |
| CIA-113 | shogun: Soeren Sonnenburg master * rd4a6596 / (6 files): fix lua examples - http://git.io/6M-E-g | 17:00 |
| nicococo | looking forward to | 17:00 |
| n4nd0 | nicococo: have you used mosek before? | 17:00 |
| nicococo | yep | 17:00 |
| nicococo | only matlab interface | 17:01 |
| n4nd0 | nicococo: ok, I am fairly new to it though | 17:01 |
| n4nd0 | nicococo: did you use matlab interface to solve this same problem? | 17:01 |
| nicococo | yes, indeed ;) | 17:01 |
| n4nd0 | nicococo: it would be awesome if you can share that code | 17:02 |
| n4nd0 | is there any problem with that? | 17:02 |
| nicococo | mmhh.. i guess i can sent it to you via email | 17:03 |
| @sonney2k | nice way to compare results... | 17:03 |
| n4nd0 | nicococo: cool | 17:03 |
| n4nd0 | nicococo: ok I have enough info to start working on this with more possibilities now :) | 17:04 |
| n4nd0 | thank you very much! | 17:04 |
| nicococo | i will send the code to you and we can chat on monday again? (or earlier if theere are problems) | 17:05 |
| n4nd0 | nicococo: I have my exam on Wednesday so let me better mail to you and see how I am doing with it ok? | 17:06 |
| nicococo | alright ;) | 17:06 |
| nicococo | bye and good luck | 17:06 |
| n4nd0 | bye! | 17:06 |
| -!- nicococo [~nico@lacedcoffee.ml.tu-berlin.de] has left #shogun [] | 17:06 | |
| @sonney2k | n4nd0, last one then? | 17:08 |
| n4nd0 | sonney2k: yes | 17:08 |
| n4nd0 | sonney2k: last one this year | 17:09 |
| @sonney2k | which subject? | 17:09 |
| @sonney2k | very good | 17:09 |
| n4nd0 | software engineering | 17:09 |
| n4nd0 | it is very boring ... I am not studying that much :( | 17:09 |
| n4nd0 | way too much theory | 17:09 |
| CIA-113 | shogun: Soeren Sonnenburg master * r29b7613 / examples/undocumented/java_modular/classifier_libsvm_minimal_modular.java : fix minimal libsvm java_modular example - http://git.io/KklDHw | 17:12 |
| CIA-113 | shogun: Soeren Sonnenburg master * r1ff3824 / examples/undocumented/csharp_modular/classifier_libsvm_minimal_modular.cs : fix minimal libsvm csharp example - http://git.io/jVvHIQ | 17:12 |
| CIA-113 | shogun: Soeren Sonnenburg master * r2be463b / (4 files): fix label get_name() - http://git.io/PZOaTw | 17:12 |
| @sonney2k | gtg | 17:12 |
| -!- n4nd0 [~nando@n155-p45.kthopen.kth.se] has quit [Quit: leaving] | 17:13 | |
| -!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has joined #shogun | 17:42 | |
| -!- gsomix [~gsomix@83.234.54.21] has joined #shogun | 18:01 | |
| gsomix | hi all | 18:01 |
| puffin444 | hey | 18:07 |
| puffin444 | how is it going? | 18:07 |
| puffin444 | gsomix: how's your first week going? | 18:09 |
| gsomix | puffin444, hey. exams :) | 18:10 |
| puffin444 | gsomix: Well that's not fun :) Are these the final exams? Does your university run through the summer? | 18:10 |
| gsomix | but I'm working with director classes... slowly | 18:11 |
| puffin444 | okay | 18:12 |
| gsomix | puffin444, not final, but most hard. I have exams until the end of June. | 18:12 |
| puffin444 | gsomix: I see. | 18:12 |
| -!- karlnapf [~heiko@host86-182-165-32.range86-182.btcentralplus.com] has joined #shogun | 18:28 | |
| -!- karlnapf1 [~heiko@host86-177-183-14.range86-177.btcentralplus.com] has quit [Ping timeout: 260 seconds] | 18:29 | |
| -!- karlnapf [~heiko@host86-182-165-32.range86-182.btcentralplus.com] has quit [Ping timeout: 260 seconds] | 18:33 | |
| @sonney2k | gsomix, where was your blog again? | 18:58 |
| * sonney2k is lost | 18:58 | |
| @sonney2k | did karlnapf have one? | 18:58 |
| @sonney2k | puffin444, do you have a blog | 19:03 |
| -!- blackburn [~blackburn@188.122.250.167] has joined #shogun | 19:06 | |
| puffin444 | sonney2k, no, I do not. | 19:12 |
| blackburn | hey there | 19:13 |
| puffin444 | hey | 19:13 |
| @sonney2k | puffin444, ok | 19:16 |
| @sonney2k | blackburn, do you remember whether karlnapf has a blog | 19:16 |
| blackburn | sonney2k: | 19:20 |
| blackburn | yes | 19:20 |
| blackburn | sonney2k: most of python modular fail | 19:20 |
| blackburn | is it the same on your machine | 19:20 |
| blackburn | sonney2k: http://herrstrathmann.de/ | 19:21 |
| blackburn | Traceback (most recent call last): | 19:23 |
| blackburn | File "classifier_averaged_perceptron_modular.py", line 32, in <module> | 19:23 |
| blackburn | classifier_averaged_perceptron_modular(*parameter_list[0]) | 19:23 |
| blackburn | File "classifier_averaged_perceptron_modular.py", line 26, in classifier_averaged_perceptron_modular | 19:23 |
| blackburn | out_labels = perceptron.apply().get_labels() | 19:23 |
| blackburn | TypeError: LinearMachine_apply expected 1 arguments, got 0 | 19:23 |
| blackburn | be back in 30 mins | 19:23 |
| @sonney2k | blackburn, ahh yes that is him :) | 19:26 |
| gsomix | sonney2k, I don't have blog. :( | 19:42 |
| gsomix | sonney2k, I will give the results for director classes tomorrow. | 19:45 |
| gsomix | but I cannot be at meeting. labs, labs, labs. last labs =____= | 19:46 |
| gsomix | I'm very sorry for that. | 19:46 |
| @sonney2k | hmmhh too bad :( | 19:50 |
| @sonney2k | blackburn, yeah same here | 19:51 |
| @sonney2k | blackburn, I finally understand the error though - but no idea how to fix it currently | 19:51 |
| gsomix | sonney2k, crazy russians universities :( | 19:52 |
| blackburn | sonney2k: well one thing I think is wrong is rename for apply() | 20:04 |
| blackburn | all machines have apply() and apply(CFeatures* data) | 20:04 |
| blackburn | however i tried that and it didn't help | 20:04 |
| blackburn | but I still believe it is something wrong | 20:04 |
| CIA-113 | shogun: Soeren Sonnenburg master * r168604a / src/interfaces/modular/Machine.i : fix %renames to not remove %extend'ed applies - http://git.io/YldzpA | 20:12 |
| @sonney2k | blackburn, please check if you multiclass liblinear works now | 20:13 |
| @sonney2k | I mean x.apply_multiclass() gives same result as x.apply() | 20:13 |
| blackburn | ok | 20:13 |
| blackburn | sonney2k: does apply work now? | 20:13 |
| blackburn | yes | 20:17 |
| @sonney2k | very good | 20:17 |
| * sonney2k is off for a swim | 20:17 | |
| blackburn | sonney2k: yes all is ok now | 20:18 |
| blackburn | lets check.sh | 20:18 |
| blackburn | I am impressed | 20:18 |
| blackburn | battleshogun operational | 20:19 |
| blackburn | ok I think we need apply_structured | 20:21 |
| blackburn | fits perfectly | 20:21 |
| -!- imeanwow [46e7e626@gateway/web/freenode/ip.70.231.230.38] has quit [Ping timeout: 245 seconds] | 20:26 | |
| -!- vikram360 [~vikram360@117.192.181.10] has quit [Read error: Connection reset by peer] | 20:27 | |
| -!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Ping timeout: 245 seconds] | 20:27 | |
| -!- vikram360 [~vikram360@117.192.181.10] has joined #shogun | 20:27 | |
| -!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has joined #shogun | 20:36 | |
| @sonney2k | all good, fresh and relaxed I am back :) | 21:19 |
| @sonney2k | blackburn, we have to make a decision about static interfaces | 21:19 |
| puffin444 | by the way, I got to present at Google Chicago last night. | 21:19 |
| blackburn | sonney2k: which decision? | 21:19 |
| @sonney2k | currently the command there is set_labels', 'TRAIN', lab) | 21:20 |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:20 | |
| blackburn | puffin444: and how it was? | 21:20 |
| puffin444 | blackburn: It was fun. The talks only lasted five minutes so I didn't get to say much | 21:20 |
| @sonney2k | now we need to specify the label type | 21:20 |
| @sonney2k | blackburn, question is how do we do it | 21:20 |
| @sonney2k | I mean we have to break the interface I am afraid | 21:21 |
| blackburn | sonney2k: automagically! | 21:21 |
| @sonney2k | blackburn, can we do that? | 21:21 |
| blackburn | sonney2k: why not - just add some function that dispatches: if (-1,+1) then create binary | 21:21 |
| blackburn | else regression | 21:22 |
| blackburn | orr? | 21:22 |
| @sonney2k | blackburn, I didn't see the beginning | 21:22 |
| blackburn | do we support multiclass at all? | 21:22 |
| @sonney2k | got disconnected ... | 21:22 |
| @sonney2k | yes | 21:22 |
| blackburn | sonney2k: we can add function there | 21:22 |
| blackburn | that dispatches | 21:22 |
| blackburn | if {-1,1} then binary | 21:22 |
| blackburn | if {0,1,..} then multiclass | 21:22 |
| blackburn | else regression | 21:22 |
| @sonney2k | ahh ok... I thought we do it more clever | 21:23 |
| blackburn | how? | 21:23 |
| blackburn | ok we can add some factory actually | 21:23 |
| @sonney2k | by checking if the machine needs a certain label type | 21:23 |
| @sonney2k | we create the label only then | 21:23 |
| blackburn | no that's too complicated I think | 21:23 |
| @sonney2k | with correct type | 21:23 |
| blackburn | however can be done too | 21:24 |
| blackburn | I like both | 21:24 |
| blackburn | puffin444: so do they liked shogun? :D | 21:24 |
| blackburn | n4nd0: what do you think about apply_structured | 21:25 |
| n4nd0 | blackburn: yeah, it should be done like that indeed | 21:25 |
| -!- vikram360 [~vikram360@117.192.181.10] has quit [Ping timeout: 250 seconds] | 21:25 | |
| puffin444 | blackburn: Yes, they were impressed by such an international team (We have 7+ countries represented) | 21:25 |
| blackburn | hmm really? | 21:25 |
| n4nd0 | blackburn: to follow the same structure there | 21:26 |
| n4nd0 | puffin444: oh true, how did you presentation go? | 21:26 |
| n4nd0 | your* | 21:26 |
| puffin444 | n4nd0: It went well. I was actually able to sort of explain generally GPs in only five mins | 21:26 |
| n4nd0 | nice | 21:27 |
| puffin444 | Their office in Chicago has a beautiful view | 21:27 |
| @sonney2k | puffin444, do you have the slides somewhere? | 21:27 |
| @sonney2k | blackburn, there is one problem | 21:27 |
| @sonney2k | we have a labels load function | 21:28 |
| puffin444 | sonney2k: Yes, I do. | 21:28 |
| puffin444 | sonne2k: All four of them :) | 21:28 |
| blackburn | load like load from file? | 21:28 |
| @sonney2k | so we need the label type before | 21:28 |
| @sonney2k | blackburn, yes | 21:28 |
| @sonney2k | puffin444, care to share? | 21:28 |
| puffin444 | sonney2k: sure, Do I put them on the mailinglist or somehow transfer them through irc? | 21:29 |
| @sonney2k | puffin444, some official post on the mailinglist would be nice - but you can also put them somewhere and link to them | 21:29 |
| @sonney2k | puffin444, or even blog about it on shogun planet :D | 21:29 |
| blackburn | I actually would like to present shogun somewhere too :D | 21:30 |
| blackburn | (okay I admit it can be an easy way to visit some country for free) | 21:30 |
| blackburn | :D | 21:30 |
| puffin444 | sonney2k: I could try blogging about it. How do I get subscription access to shogun planet? | 21:36 |
| @sonney2k | puffin444, create a blog/rss feed somewhere and tell me the feed url | 21:39 |
| @sonney2k | blackburn, most certainly welcome :D | 21:39 |
| blackburn | sonney2k: yeah just need to find some conference I think? | 21:44 |
| @sonney2k | and get travel money from uni or so | 21:45 |
| @sonney2k | blackburn, my proposal wrt loading labels is | 21:47 |
| @sonney2k | I load regression labels | 21:47 |
| @sonney2k | and then we do the automagic conversion again | 21:48 |
| blackburn | hmm yes makes sense | 21:50 |
| -!- ckwidmer [~chris@HSI-KBW-046-005-237-106.hsi8.kabel-badenwuerttemberg.de] has joined #shogun | 21:56 | |
| @sonney2k | the chris | 21:57 |
| ckwidmer | hi all | 21:57 |
| blackburn | oh | 21:57 |
| puffin444 | hey chris | 21:57 |
| puffin444 | eating lunch. be back later. | 21:57 |
| -!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Quit: Page closed] | 21:58 | |
| @sonney2k | blackburn, there is an issue with the automagic label stuff | 21:59 |
| @sonney2k | currently our exmaples use +1/-1 labels for regression | 22:00 |
| @sonney2k | blackburn, so what do we do now? | 22:03 |
| @sonney2k | create examples with real valued labels? | 22:04 |
| blackburn | sonney2k: hmm or act in the way you suggested | 22:04 |
| blackburn | depending on machine | 22:04 |
| -!- ckwidmer [~chris@HSI-KBW-046-005-237-106.hsi8.kabel-badenwuerttemberg.de] has quit [Remote host closed the connection] | 22:18 | |
| @sonney2k | blackburn, any idea how to add anything to data? | 22:26 |
| @sonney2k | I did add regression labels | 22:26 |
| blackburn | sonney2k: to data like? | 22:26 |
| @sonney2k | but git push sais everything up-to-date | 22:27 |
| blackburn | ah shogun-data? | 22:27 |
| @sonney2k | y | 22:27 |
| blackburn | sonney2k: hmm probably you didn't 'activate' it as git repo | 22:28 |
| blackburn | I tried to remove toy here | 22:29 |
| blackburn | sonney2k: git submodule update --init? | 22:30 |
| CIA-113 | shogun: Soeren Sonnenburg master * r99a35be / (13 files in 6 dirs): fix static interfaces - http://git.io/TYtpIQ | 22:32 |
| @sonney2k | blackburn, no I mean I want to add the data set | 22:34 |
| @sonney2k | commit & push it to shogun-data | 22:34 |
| @sonney2k | but it claims that this repo is up-to-date | 22:34 |
| @sonney2k | even though I have the new data set in there now | 22:34 |
| blackburn | sonney2k: yes - because it considers data as main git | 22:34 |
| shogun-buildbot | build #990 of libshogun is complete: Failure [failed git] Build details are at http://www.shogun-toolbox.org/buildbot/builders/libshogun/builds/990 blamelist: sonne@debian.org | 22:35 |
| blackburn | is it fresh clone? did you set up submodules? | 22:35 |
| @sonney2k | blackburn, I go to shogun/data | 22:35 |
| @sonney2k | how do I convince git that this is not part of shogun/ ? | 22:36 |
| blackburn | sonney2k: yes but did you 'git submodule update --init' in shogun/? | 22:36 |
| blackburn | or somehow | 22:36 |
| @sonney2k | blackburn, yeah to fetch the data | 22:36 |
| @sonney2k | then I modified the data | 22:37 |
| @sonney2k | committed | 22:37 |
| blackburn | hmm that's strange then | 22:37 |
| @sonney2k | added a new remote | 22:37 |
| @sonney2k | and push now says f... you | 22:37 |
| blackburn | hmm | 22:38 |
| blackburn | sonney2k: one idea | 22:38 |
| blackburn | sonney2k: git status? | 22:38 |
| blackburn | # Not currently on any branch. | 22:38 |
| blackburn | right? | 22:38 |
| blackburn | save your data and recommit if so | 22:39 |
| @sonney2k | yeah indeed | 22:41 |
| CIA-113 | shogun: Soeren Sonnenburg master * r23dfd86 / data : add new data - http://git.io/p60e2w | 22:44 |
| blackburn | sonney2k: interesting fact it all goes to valhalla if not on branch | 22:44 |
| @sonney2k | blackburn, yeah sometimes git is dangerous | 22:44 |
| CIA-113 | shogun: Soeren Sonnenburg master * r7866158 / (2 files): forgot .dat suffix - http://git.io/71UatQ | 22:51 |
| blackburn | sonney2k: I am afraid you put labels creation into the loop | 22:52 |
| blackburn | ah right | 22:52 |
| blackburn | break | 22:52 |
| shogun-buildbot | build #991 of libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/libshogun/builds/991 | 22:53 |
| @sonney2k | blackburn, heh | 22:53 |
| @sonney2k | clever me :D | 22:53 |
| blackburn | yeah | 22:53 |
| blackburn | sonney2k: any idea how to boost accuracy for 0.5-1%? | 22:54 |
| blackburn | multiclass, you know | 22:54 |
| @sonney2k | gsomix, in case you get the director kernels to work there is another cool python feature (python buffer protocol) that we all want :) | 22:54 |
| @sonney2k | blackburn, what? | 22:54 |
| blackburn | sonney2k: I desperately need to get 0.98 instead of 0.97 :D | 22:55 |
| blackburn | maybe some tricks? | 22:55 |
| @sonney2k | no clue | 22:55 |
| gsomix | sonney2k, aha, aha. | 22:56 |
| @sonney2k | blackburn, could you fix these warnings http://shogun-toolbox.org/buildbot/builders/libshogun/builds/991/steps/compile/logs/warnings%20%282%29 | 22:56 |
| blackburn | yeap | 22:56 |
| blackburn | sonney2k: any idea why liblinear computing time -> infty after prunevarsubmean? | 22:56 |
| @sonney2k | no | 22:57 |
| shogun-buildbot | build #893 of cmdline_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/893 | 22:57 |
| blackburn | hmm my issues are all strange | 22:57 |
| @sonney2k | gsomix, ahh and I think we need you to turn on the tests afterwards ... | 23:03 |
| blackburn | sonney2k: which normalization is better for linear SVM? 1/2? | 23:06 |
| gsomix | sonney2k, which tests? | 23:07 |
| shogun-buildbot | build #782 of octave_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/782 | 23:08 |
| @sonney2k | blackburn, depends on problem sqrtdiag quite often | 23:11 |
| @sonney2k | but it also helps to use binneddotfeatures | 23:12 |
| blackburn | sonney2k: what is advantage of using binned features? | 23:12 |
| @sonney2k | you can learn piece wise linear functions | 23:12 |
| @sonney2k | so with enough bins basically any function | 23:13 |
| @sonney2k | works pretty well | 23:13 |
| @sonney2k | and no normalization issues | 23:13 |
| shogun-buildbot | build #859 of python_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/859 | 23:13 |
| blackburn | sonney2k: with which features? | 23:13 |
| blackburn | anything special? | 23:13 |
| -!- PolarBearAttack [46e7e626@gateway/web/freenode/ip.70.231.230.38] has joined #shogun | 23:13 | |
| PolarBearAttack | Sup | 23:13 |
| PolarBearAttack | What is shotgun toolbox about? | 23:13 |
| blackburn | shotgun :D | 23:13 |
| PolarBearAttack | shogun, sorry | 23:14 |
| blackburn | I'd say its name is shogun | 23:14 |
| blackburn | :) | 23:14 |
| blackburn | machine learning | 23:14 |
| @sonney2k | world domination | 23:14 |
| PolarBearAttack | benefits? | 23:14 |
| @sonney2k | ego trips | 23:14 |
| PolarBearAttack | How do you acquire such practice? | 23:14 |
| n4nd0 | :D | 23:14 |
| PolarBearAttack | requirements? how do you know | 23:14 |
| blackburn | PolarBearAttack: for example I use it to construct road sign recognition 'system' | 23:15 |
| -!- PolarBearAttack was kicked from #shogun by sonney2k [PolarBearAttack] | 23:15 | |
| -!- PolarBearAttack [46e7e626@gateway/web/freenode/ip.70.231.230.38] has joined #shogun | 23:15 | |
| PolarBearAttack | why | 23:15 |
| blackburn | why :) | 23:15 |
| @sonney2k | how do we disable that bot? | 23:15 |
| @sonney2k | blackburn, is that your grisly? | 23:15 |
| PolarBearAttack | let me be part of the society? | 23:15 |
| blackburn | hmm I did not recognize it as bot :D | 23:15 |
| blackburn | PolarBearAttack: are you? | 23:16 |
| n4nd0 | sonney2k: it might dangerous to kick polar bears :D | 23:16 |
| @sonney2k | true | 23:16 |
| @sonney2k | they can bite | 23:16 |
| @sonney2k | PolarBearAttack, can you? | 23:16 |
| PolarBearAttack | I'm coherent, I will contribute | 23:16 |
| blackburn | sonney2k: can you please describe how can I appoach anything with binned features? | 23:17 |
| @sonney2k | blackburn, are you coherent and will contribute? | 23:18 |
| blackburn | sonney2k: I promise I will | 23:18 |
| @sonney2k | PolarBearAttack, see | 23:18 |
| @sonney2k | blackburn, has a grisly at his home | 23:18 |
| @sonney2k | we are not afraid... | 23:19 |
| @sonney2k | blackburn, you have to give it the bins | 23:19 |
| PolarBearAttack | Can people talk in terms of mathematics? | 23:19 |
| blackburn | PolarBearAttack: f(x) | 23:19 |
| @sonney2k | there is an example | 23:19 |
| @sonney2k | have a look | 23:20 |
| blackburn | sonney2k: bins like [0,1] (1,2] | 23:20 |
| blackburn | right? | 23:20 |
| @sonney2k | I think so | 23:20 |
| @sonney2k | blackburn, how many dims do you have? | 23:20 |
| blackburn | sonney2k: depends how many hog cells I have but generally varying from 200 to 900 | 23:20 |
| @sonney2k | PolarBearAttack, f(x)=\sum_i \alpha_i y_i K(x_i, x) | 23:20 |
| @sonney2k | blackburn, so you can easily use 100 bins per feature | 23:21 |
| @sonney2k | blackburn, how many examples? | 23:21 |
| PolarBearAttack | sonney2k: , how did you formulate that thought? | 23:21 |
| blackburn | sonney2k: 39K | 23:21 |
| @sonney2k | not so many | 23:21 |
| blackburn | hmm I could generate some with rotations, etc :D | 23:21 |
| @sonney2k | PolarBearAttack, I used a brain computer interface | 23:22 |
| -!- mode/#shogun [+o blackburn] by sonney2k | 23:22 | |
| -!- mode/#shogun [-o blackburn] by ChanServ | 23:22 | |
| blackburn | argh | 23:22 |
| PolarBearAttack | did you subvocalized? | 23:22 |
| blackburn | not registered | 23:22 |
| PolarBearAttack | Formulating thought in terms of symbol is pretty powerful | 23:23 |
| n4nd0 | this is getting awkward ... | 23:23 |
| blackburn | PolarBearAttack: all the machine learning can be formulated as searching f such that f: X->Y! | 23:23 |
| -!- mode/#shogun [+b *!*46e7e626@*.70.231.230.38] by sonney2k | 23:23 | |
| -!- PolarBearAttack was kicked from #shogun by sonney2k [PolarBearAttack] | 23:23 | |
| blackburn | why that was pretty funny | 23:23 |
| @sonney2k | black /query him | 23:25 |
| @sonney2k | if you want to talk to him/it | 23:25 |
| blackburn | sonney2k: argh I forgot with HKM I get *3 features | 23:25 |
| @sonney2k | bed time! | 23:25 |
| blackburn | damn | 23:25 |
| n4nd0 | sonney2k: gute Nacht | 23:26 |
| shogun-buildbot | build #873 of r_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/873 | 23:27 |
| gsomix | good night guys | 23:37 |
| n4nd0 | good night gsomix | 23:39 |
| shogun-buildbot | build #563 of csharp_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/csharp_modular/builds/563 blamelist: sonne@debian.org | 23:56 |
| --- Log closed Fri May 25 00:00:41 2012 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!