| --- Log opened Thu Aug 02 00:00:01 2012 | ||
| --- Day changed Thu Aug 02 2012 | ||
| n4nd0 | and MOSEK libraries | 00:00 |
|---|---|---|
| blackburn | okay sure but why not to have it automatically? | 00:00 |
| n4nd0 | no idea | 00:00 |
| blackburn | so should I do that? :) | 00:00 |
| n4nd0 | maybe Soeren just didn't want that I'd spend time with that | 00:00 |
| blackburn | yeah I can | 00:00 |
| n4nd0 | sure, why not | 00:00 |
| blackburn | n4nd0: can you help me with that a little? | 00:00 |
| n4nd0 | blackburn: yes | 00:01 |
| blackburn | i need a minimal example | 00:01 |
| blackburn | that fails w/o mosek | 00:01 |
| blackburn | and flags you use | 00:01 |
| blackburn | -lmosek or so? | 00:01 |
| n4nd0 | yes | 00:01 |
| n4nd0 | I can give you that | 00:01 |
| n4nd0 | I will do that tomorrow morning | 00:01 |
| n4nd0 | I am going early today to sleep | 00:02 |
| blackburn | wanted to make it now :) | 00:02 |
| blackburn | just say what are flags | 00:02 |
| blackburn | -lmosek? | 00:02 |
| n4nd0 | wait a moment, I sent Nico a mail with the exact steps | 00:03 |
| n4nd0 | -- In the line that start with DEFINES add -DUSE_MOSEK. | 00:03 |
| n4nd0 | -- In the line INCLUDES add -I/usr/local/etc/mosek/6/tools/platform/linux32x86/h | 00:03 |
| n4nd0 | where /usr/local/etc/ is the path to your mosek installation directory that may | 00:03 |
| n4nd0 | be different for you. | 00:03 |
| n4nd0 | -- In the line LINKFLAGS add -L/usr/local/etc/mosek/6/tools/platform/linux32x86/bin | 00:03 |
| n4nd0 | where it applies the same as before about the mosek isntallation directory. | 00:04 |
| n4nd0 | - In POSTLINKFLAGS add -lmosek. | 00:04 |
| n4nd0 | I love the function of search in the mails :D | 00:04 |
| blackburn | thanks got it | 00:04 |
| blackburn | n4nd0: did you go already? | 00:12 |
| n4nd0 | blackburn: no | 00:14 |
| n4nd0 | tell me | 00:14 |
| blackburn | n4nd0: these paths wouldn't work as something default I am afraid | 00:15 |
| blackburn | so one would need to set up these paths on configure | 00:15 |
| n4nd0 | whereis mosek outputs /usr/local/etc/mosek in my machine | 00:16 |
| n4nd0 | maybe that could work? | 00:16 |
| n4nd0 | although I do remember I told Soeren the same thing and he said that it was not platform independent | 00:16 |
| blackburn | does it have pkg-config? | 00:16 |
| n4nd0 | no idea what that is | 00:16 |
| blackburn | cool thing allowing to do | 00:16 |
| blackburn | something like | 00:16 |
| blackburn | pkg-config --cflags mosek | 00:17 |
| n4nd0 | let me try to exec that | 00:17 |
| blackburn | but probably it would not work | 00:17 |
| n4nd0 | Package mosek was not found in the pkg-config search path. | 00:17 |
| n4nd0 | Perhaps you should add the directory containing `mosek.pc' | 00:17 |
| n4nd0 | to the PKG_CONFIG_PATH environment variable | 00:17 |
| n4nd0 | No package 'mosek' foun | 00:17 |
| n4nd0 | you were right | 00:17 |
| n4nd0 | doesn't work | 00:17 |
| n4nd0 | :S | 00:17 |
| blackburn | okay I've added something | 00:19 |
| CIA-18 | shogun: Sergey Lisitsyn master * ref83b92 / src/configure : Added mosek detection - https://github.com/shogun-toolbox/shogun/commit/ef83b921566fc6269a33842578c6097c72932908 | 00:19 |
| shogun-buildbot | build #240 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/240 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 00:19 |
| blackburn | ehmm okay | 00:21 |
| n4nd0 | :) | 00:22 |
| n4nd0 | ok good night now then! | 00:40 |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 00:40 | |
| -!- needsch [~user@ip-176-198-229-75.unitymediagroup.de] has quit [Quit: Verlassend] | 00:48 | |
| shogun-buildbot | build #241 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/241 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 00:50 |
| CIA-18 | shogun: Sergey Lisitsyn master * rbe2d76b / (2 files): Fixed apply method call of multitask logistic regression - https://github.com/shogun-toolbox/shogun/commit/be2d76b8b4b98ba7bfe8e05adcad468565904138 | 01:23 |
| -!- blackburn [~blackburn@109.226.100.87] has quit [Quit: Leaving.] | 01:25 | |
| shogun-buildbot | build #242 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/242 | 02:05 |
| -!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 260 seconds] | 02:09 | |
| -!- av3ngr [av3ngr@nat/redhat/x-cwemmkxkyzgxvooo] has joined #shogun | 03:24 | |
| -!- av3ngr is now known as abn | 03:24 | |
| -!- puffin444 [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Quit: Page closed] | 03:44 | |
| shogun-buildbot | build #47 of nightly_default is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/47 | 03:59 |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 06:53 | |
| -!- abn [av3ngr@nat/redhat/x-cwemmkxkyzgxvooo] has quit [Quit: That's all folks!] | 07:41 | |
| -!- uricamic [~uricamic@2001:718:2:1634:a48f:d0e6:2584:7cff] has joined #shogun | 08:56 | |
| -!- cronor [~cronor@fb.ml.tu-berlin.de] has joined #shogun | 09:34 | |
| -!- cronor [~cronor@fb.ml.tu-berlin.de] has left #shogun [] | 09:34 | |
| -!- flxb [~cronor@fb.ml.tu-berlin.de] has joined #shogun | 09:34 | |
| -!- pluskid [~pluskid@108.171.196.83] has joined #shogun | 09:35 | |
| -!- Marina [82955874@gateway/web/freenode/ip.130.149.88.116] has joined #shogun | 09:35 | |
| Marina | hello, if i train a svm with dna sequence for splice site recognition, do the sequences with the splice site must have the +1 or the -1 label? | 09:43 |
| n4nd0 | hi Marina | 09:49 |
| n4nd0 | yes, if you use a binary classifier, for example one of the binary SVMs, the labels have to be either +1 or -1 | 09:50 |
| Marina | hi n4nd0 so, there is no difference, if i use the +1 one label for the sequences with the splice site, i.e. the positive examples, or if i use the -1 label for the sequences with the splice site? | 10:00 |
| n4nd0 | Marina: what are the two classes in your problem? I am not familiarized with splice site recognition | 10:01 |
| n4nd0 | but I'd say that in any binary classification problem you should learn the same model if the labels are swapped | 10:03 |
| n4nd0 | so it shouldn't make a difference | 10:03 |
| Marina | after training the svm i make a poim with the svm result - and the picture look different when training with -1 or +1 for the splice site sequences. | 10:07 |
| n4nd0 | Marina: even if they look different, aren't they equivalent in some sense? | 10:10 |
| n4nd0 | Marina: have you tried comparing the results of the weight vector after training with +1 and -1? | 10:11 |
| Marina | i will proof it! Thanks a lot | 10:16 |
| n4nd0 | you are welcome | 10:16 |
| n4nd0 | Marina: what svm classifier are you using by the way? | 10:16 |
| Marina | i use LibSVM | 10:24 |
| n4nd0 | ok | 10:26 |
| n4nd0 | I have just a test with libsvm | 10:26 |
| n4nd0 | I have train my svm two times, the second time swapping the labels of the training examples with respect to the first time | 10:27 |
| -!- pluskid [~pluskid@108.171.196.83] has quit [Quit: Leaving] | 10:28 | |
| n4nd0 | and I have printed the alphas, which define the model learnt, both times | 10:28 |
| n4nd0 | the values are the same in both cases, with the exception that the signs are changed | 10:28 |
| n4nd0 | e.g. the first time I get something like [0.68 -1 ...], the second time [-0.68 1 ...] | 10:29 |
| n4nd0 | and that actually makes sense, graphically the hyperplane is the same | 10:29 |
| n4nd0 | but the meaning of each region changes when the labels are swapped | 10:30 |
| n4nd0 | Marina: could it be that the change in the signs made the poim different? | 10:31 |
| koen-shogun | by default, the costs (e.g. weights) of the classes are the same | 11:03 |
| koen-shogun | so you'd expect nearly the same solution | 11:03 |
| Marina | n4nd0 yes - that should be the reason! | 11:03 |
| -!- Netsplit *.net <-> *.split quits: audy | 11:15 | |
| -!- Netsplit over, joins: audy | 11:17 | |
| -!- blackburn [~blackburn@109.226.100.87] has joined #shogun | 12:01 | |
| -!- blackburn [~blackburn@109.226.100.87] has quit [Quit: Leaving.] | 12:34 | |
| -!- Marina [82955874@gateway/web/freenode/ip.130.149.88.116] has quit [Ping timeout: 245 seconds] | 13:36 | |
| -!- koen-shogun [~ksande@195.114.233.130] has left #shogun [] | 14:50 | |
| -!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun | 15:17 | |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 15:33 | |
| -!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Quit: Page closed] | 15:39 | |
| -!- blackburn [~blackburn@109.226.100.87] has joined #shogun | 16:03 | |
| -!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun | 16:30 | |
| yoo | hi all | 16:30 |
| blackburn | hey | 16:36 |
| yoo | blackburn: reading your mail, thx for answering | 16:42 |
| blackburn | yoo: btw why do you use C++? | 16:44 |
| yoo | blackburn: simply because I am more comfortable with c++ | 16:48 |
| yoo | blackburn: and since I have made some modifications in c++ shogun code .. | 16:49 |
| yoo | blackburn:do you usually use python interface ? | 16:49 |
| blackburn | yeah | 16:49 |
| -!- audy is now known as octopine | 16:49 | |
| blackburn | well it is just the same actually and loading data in C++ is usually a kind of pain | 16:50 |
| yoo | the thing is when I create my own c++ class, I have no experience in wrapping in python form | 16:51 |
| yoo | could be the time I get into this .. | 16:51 |
| -!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Quit: Page closed] | 17:05 | |
| blackburn | hmm I am unsure what do you mean | 17:08 |
| -!- uricamic [~uricamic@2001:718:2:1634:a48f:d0e6:2584:7cff] has quit [Quit: Leaving.] | 17:10 | |
| -!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun | 17:33 | |
| -!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Quit: Page closed] | 18:05 | |
| -!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has joined #shogun | 18:05 | |
| KMcQuisten | Hello all | 18:05 |
| blackburn | hello | 18:07 |
| KMcQuisten | I'm getting some really odd behavior form LibSVR, and I was wondering if you all could help. I can set up a kernel, initialize it, load that kernel and some labels into the LibSVR object, and then train the machine. However, when I try and make a prediction, either on the training set using apply() or on a set of unseen features using apply(Cfeatures), it predicts an output of 1.0 for everythign, regardless of what I input. | 18:10 |
| KMcQuisten | I have a script I can send you to reproduce this. | 18:11 |
| blackburn | hmm sure, show me | 18:12 |
| KMcQuisten | Sent | 18:16 |
| KMcQuisten | What do you think? | 18:22 |
| blackburn | give me 5 mins more :) | 18:24 |
| KMcQuisten | No problem :) | 18:24 |
| -!- heiko [~heiko@host86-181-156-213.range86-181.btcentralplus.com] has joined #shogun | 18:27 | |
| blackburn | ahh | 18:36 |
| KMcQuisten | So what's the verdict? | 18:37 |
| blackburn | KMcQuisten: for now use apply_regression | 18:37 |
| blackburn | instead of apply | 18:37 |
| KMcQuisten | Allrighty :) | 18:37 |
| blackburn | I will check why it calls apply_binary here | 18:38 |
| KMcQuisten | Thanks for the help | 18:39 |
| blackburn | heiko! | 18:41 |
| heiko | blackburn | 18:41 |
| heiko | hi! | 18:41 |
| blackburn | hey | 18:41 |
| blackburn | a few things to discuss I have you to | 18:42 |
| heiko | ohoh | 18:42 |
| blackburn | heiko: mind talk I yoda like? | 18:42 |
| blackburn | :D | 18:42 |
| heiko | thou shall he | 18:42 |
| heiko | :D | 18:42 |
| blackburn | okay | 18:43 |
| blackburn | I heard you like multiclass | 18:43 |
| blackburn | so I put multiclass into multiclass | 18:43 |
| blackburn | okay I stop joking | 18:43 |
| blackburn | heiko: first thing | 18:43 |
| heiko | go ahead! | 18:44 |
| blackburn | are you familiar how multiclass machines work currently? | 18:45 |
| blackburn | you must be | 18:45 |
| blackburn | :) | 18:45 |
| heiko | lets just say yes | 18:45 |
| heiko | I fixed a few bugs in these so almost | 18:46 |
| blackburn | how do you think we could pass | 18:46 |
| blackburn | these scores | 18:46 |
| blackburn | to some evaluation | 18:46 |
| blackburn | problem is that | 18:46 |
| blackburn | multiclass labels is num + confidence | 18:46 |
| heiko | you mean the score from OvO etc? | 18:47 |
| blackburn | but for multiclass ROC we would need more confidences | 18:47 |
| blackburn | OvO ROC is probably not really informativ | 18:47 |
| blackburn | e | 18:47 |
| heiko | mmh | 18:47 |
| heiko | I see | 18:47 |
| blackburn | but OvR one would be nice | 18:47 |
| heiko | we got confidences on the one side | 18:47 |
| heiko | and scores on the other | 18:47 |
| blackburn | yeah | 18:47 |
| heiko | As I see it we are currently abusing the term confidence by putting scores in there that are not calibrated probabilities right? | 18:48 |
| heiko | so why not have two fields: scores AND probabilities | 18:48 |
| heiko | probabilities can be used as scores | 18:48 |
| heiko | but not the other way round | 18:48 |
| heiko | and also it would have the advantage that users are not confused (I tend to be confused with these things for example) | 18:49 |
| blackburn | hmm | 18:50 |
| blackburn | but wait | 18:51 |
| heiko | any problems with that? | 18:52 |
| blackburn | what do we need to compute OvR roc? | 18:52 |
| heiko | Possible with both | 18:53 |
| heiko | oh wait | 18:53 |
| heiko | no | 18:53 |
| heiko | its not possible to compute real probabilities for OvR | 18:53 |
| blackburn | so for each vector | 18:53 |
| heiko | its not really a probabilistic framework | 18:53 |
| blackburn | we need all outputs | 18:53 |
| heiko | I mean its possible to approximate this | 18:53 |
| blackburn | is OvO probabilistic? | 18:53 |
| heiko | but scores are fine for ROC are they | 18:54 |
| heiko | no | 18:54 |
| heiko | SVM have no probabilistic interpretation anyway | 18:54 |
| heiko | all probabilities are just approximations or fitted sigmoids | 18:54 |
| heiko | couldnt we make this all optional? | 18:55 |
| heiko | I mean currently shogun cannot compute any probabilities on the output | 18:55 |
| heiko | I wanted to add that anyway once the labels could do it | 18:55 |
| heiko | but I still think its a good idea to distringuish | 18:55 |
| blackburn | hmm why do you want probabilities? | 18:55 |
| heiko | for example, the GP has a probabilistic interpretation, there you have only probabilities and not really scoresy | 18:55 |
| heiko | well for multiclass its great to compute posterior probabilities isnt it? | 18:56 |
| blackburn | probably | 18:56 |
| heiko | first class 33% second class 32%, which one to take | 18:56 |
| heiko | using the one with largest scores looses something | 18:56 |
| heiko | on the other side: | 18:56 |
| heiko | whenever you have probabilities, you dont need scores anymore | 18:56 |
| heiko | its just when you have no probabilities | 18:57 |
| blackburn | yeah | 18:57 |
| heiko | then you need the scores | 18:57 |
| blackburn | problem is that | 18:57 |
| blackburn | we currently discard scores and probabilities of rest classes | 18:57 |
| blackburn | when choosing one | 18:57 |
| heiko | but its also possible to "fake" probabilities, as for example for multiclass SVMs | 18:57 |
| heiko | then you got two different numbers | 18:57 |
| heiko | ah ok | 18:57 |
| heiko | blackburn, what were the other things? | 19:03 |
| blackburn | I can't really see how to handle that | 19:03 |
| blackburn | heiko: other other | 19:03 |
| blackburn | other is | 19:03 |
| blackburn | modelselection output | 19:03 |
| blackburn | what do you think is the way to *store* these things it prints now | 19:04 |
| heiko | why do you want to do that? | 19:04 |
| heiko | for producing plots etc? | 19:04 |
| heiko | what about adding a method that collects all performance measures for all parameter combinations? | 19:05 |
| heiko | and then output and array with that | 19:05 |
| blackburn | heiko: hmm | 19:07 |
| blackburn | heiko: yes, for plots and everything | 19:07 |
| heiko | what exactly do you want to do with that? | 19:07 |
| blackburn | okay currently it prints everything | 19:07 |
| blackburn | probably it would be better to store it | 19:07 |
| blackburn | and not to print | 19:07 |
| heiko | do you also want to store all indices etc? | 19:08 |
| heiko | or just performance per point? | 19:08 |
| blackburn | no, indices | 19:08 |
| blackburn | too | 19:08 |
| blackburn | probably | 19:08 |
| heiko | why? | 19:08 |
| blackburn | I think it is needed sometimes for reproducibility | 19:09 |
| blackburn | people really often make they own xval and modelselection snippets | 19:09 |
| heiko | I cannot imagine why one would want that. The thing about x-validation is that is gives you a normal distributed estimate of the real performance | 19:10 |
| heiko | I agree, because shogun often does not work ;) | 19:10 |
| heiko | when it does, one doesnt need the indices in my eyes, or I can't imagine a use-case for that | 19:10 |
| heiko | another thought: SHOGUN's model selection/x-validation is more for end-user cases: to automagically select parameters or get a performance measure. Its not really designed for analysing what happens in x-val. | 19:14 |
| heiko | But thats a wide field and its almost impossible to cover everything that you might want to do with x-val. | 19:15 |
| heiko | so if people want to analyse more subtle stuff, they will have to write their own routines | 19:15 |
| heiko | and if they just want the distribution of the performance measure, we can add that | 19:16 |
| heiko | blackburn, what about the testing PR from wiking? | 19:17 |
| blackburn | oops | 19:17 |
| heiko | that is something I am really interested in :) | 19:17 |
| blackburn | sorry got distracted | 19:17 |
| heiko | no worries :) | 19:17 |
| blackburn | okay | 19:17 |
| blackburn | I never got back to that PR atually | 19:18 |
| blackburn | heiko: can't we put GPL to readme or so? | 19:30 |
| blackburn | it is so damn big | 19:30 |
| heiko | ehm, where? | 19:31 |
| heiko | in the tutorial? | 19:31 |
| heiko | blackburn, whats wrong with having it in the appendix? | 19:31 |
| blackburn | heiko: yeah, why to have it where? | 19:31 |
| heiko | I had this book-style in mind, so if printed, its good to put it somewhere | 19:32 |
| heiko | but I dont insist on it | 19:32 |
| heiko | we can give an url :) | 19:32 |
| heiko | just | 19:32 |
| heiko | for this license for documentation | 19:33 |
| heiko | it sais that the document contains a section where the full license is printed | 19:33 |
| heiko | thats why I added it | 19:33 |
| blackburn | I see | 19:33 |
| blackburn | okay let it stay then | 19:33 |
| heiko | but I dont know whether its compulsory | 19:33 |
| CIA-18 | shogun: Sergey Lisitsyn master * ra51e803 / src/interfaces/modular/Machine.i : Added apply rename for libsvr - https://github.com/shogun-toolbox/shogun/commit/a51e803b4b38afb74945b1542af6e286612306db | 19:34 |
| blackburn | heiko: btw there is a shogun book already :D | 19:35 |
| blackburn | http://www.amazon.co.uk/Shogun-toolbox-Ronald-Jesse-Russell/dp/B007OV760A | 19:36 |
| heiko | lol :D | 19:36 |
| heiko | is that actually written? | 19:37 |
| heiko | look fakey | 19:38 |
| blackburn | fake for sure | 19:38 |
| blackburn | heiko: did you visit something at games? | 19:39 |
| heiko | blackburn, not yet | 19:39 |
| blackburn | will you? :) | 19:39 |
| heiko | but I cannot get the motivation for this, I find it rather boring -- but I feel bad for this | 19:39 |
| heiko | and so much work currently | 19:40 |
| blackburn | heiko: probably it is not so special in your countries :D | 19:41 |
| blackburn | there is a good old hype around this | 19:41 |
| blackburn | "we must secure our regime via a lot of gold medals!!" | 19:41 |
| blackburn | that's how china acts actually and soviet union been acting :D | 19:41 |
| heiko | well, yes it is, its actually amazing, and also a great opportunity since its next door, thats why I feel bad for not being interested :D | 19:41 |
| heiko | yeah | 19:41 |
| heiko | I like that approach btw ;) | 19:41 |
| blackburn | why? | 19:42 |
| heiko | because it seems to naive | 19:42 |
| blackburn | well competition is good | 19:42 |
| heiko | kind of odd | 19:42 |
| blackburn | but with all these dopes china uses (I am sure they do) | 19:42 |
| blackburn | it stops to be a competition of athletes :D | 19:42 |
| heiko | yeah | 19:42 |
| heiko | everybody does this | 19:43 |
| heiko | the chinese are just best at staying closely under the accepted limit | 19:43 |
| blackburn | they just find it important to produce these roboathletes, normal countries are more relaxed I believe :) | 19:44 |
| -!- flxb [~cronor@fb.ml.tu-berlin.de] has quit [Ping timeout: 240 seconds] | 19:46 | |
| heiko | probably | 19:47 |
| heiko | one GSoC student is an athlete btw did you know that? | 19:47 |
| blackburn | wow | 20:05 |
| blackburn | no I didn't know | 20:05 |
| blackburn | that's crazy | 20:06 |
| -!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has quit [Quit: Page closed] | 20:22 | |
| blackburn | how does he manage to do that :D | 20:26 |
| octopine | can anyone help me understand what this dip is all about? http://d.pr/i/sYmI+#.png | 20:35 |
| blackburn | uh interesting | 20:39 |
| blackburn | heiko: may be you? | 20:40 |
| blackburn | I am out of expertise there | 20:40 |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has joined #shogun | 20:42 | |
| heiko | octopine, blackburn, sorry just read this, got to go, may be able to help later | 20:55 |
| -!- heiko [~heiko@host86-181-156-213.range86-181.btcentralplus.com] has left #shogun [] | 20:55 | |
| blackburn | sonney2k: then you :) | 20:55 |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has quit [Ping timeout: 252 seconds] | 21:19 | |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has joined #shogun | 21:20 | |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 21:47 | |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has joined #shogun | 21:48 | |
| @sonney2k | octopine, well can happen... prc curve is not convex monotone or anything... | 22:12 |
| blackburn | sonney2k: but doesn't it mean something? | 22:18 |
| blackburn | that dip | 22:18 |
| blackburn | sonney2k: locking is really helpful with multitask - that's really good we introduced it | 22:56 |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 22:59 | |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has joined #shogun | 22:59 | |
| -!- zxtx [~zv@183.sub-174-253-244.myvzw.com] has joined #shogun | 23:36 | |
| -!- zxtx [~zv@183.sub-174-253-244.myvzw.com] has quit [Read error: Connection reset by peer] | 23:44 | |
| @sonney2k | blackburn, thank heiko for that... | 23:48 |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has quit [Read error: Connection reset by peer] | 23:49 | |
| blackburn | sonney2k: do we want to have bayesian networks? | 23:49 |
| -!- cronor [~cronor@g229081189.adsl.alicedsl.de] has joined #shogun | 23:50 | |
| -!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 23:51 | |
| n4nd0 | hey blackburn | 23:53 |
| blackburn | hey n4nd0 | 23:53 |
| n4nd0 | how is it going? | 23:53 |
| blackburn | well pretty fine, finishing up yet another refactoring of multiclass | 23:54 |
| blackburn | and you? | 23:54 |
| @sonney2k | n4nd0, you are right about the label stuff - if you swap +1 <-> -1 POIMs will be inverted... | 23:54 |
| n4nd0 | I am fine too | 23:54 |
| blackburn | n4nd0: bioinformatics expert now? ;) | 23:54 |
| n4nd0 | sonney2k: hey! I didn't expect you around here now :) | 23:54 |
| n4nd0 | blackburn: haha not at all | 23:54 |
| blackburn | no matter how long I support shogun I NEVER damn understand these POIMs and stuff | 23:55 |
| n4nd0 | sonney2k: I was going to consult with blackburn, tomorrow we have a meeting right? | 23:55 |
| @sonney2k | yes | 23:55 |
| @sonney2k | blackburn, well MKL and POIMs are not exactly trivial stuff | 23:55 |
| blackburn | sonney2k: mind send us a reminder? | 23:55 |
| blackburn | :) | 23:56 |
| n4nd0 | sonney2k: unfortunately I won't be able to make it | 23:56 |
| n4nd0 | I will be in a plane to Spain tomorrow at that time | 23:56 |
| n4nd0 | I thought about sending to the mailing list an email with the summary of the state of the project, as we normally do in these meetings | 23:57 |
| blackburn | n4nd0: what you wanted to consult me about? | 23:57 |
| n4nd0 | blackburn: if there was going to be a meeting or was it just a mistake in my agenda | 23:57 |
| blackburn | ahh | 23:57 |
| @sonney2k | n4nd0, hmmhh so neither you nor nico will attend? | 23:57 |
| @sonney2k | blackburn, argh so let me send some reminder ... | 23:58 |
| n4nd0 | sonney2k: I am not sure about Nico, but he has not been answering my mails for a long while ... so my guess is that we won't make it | 23:58 |
| n4nd0 | I am sorry | 23:58 |
| blackburn | I bet it will be low population | 23:58 |
| n4nd0 | sonney2k: if you let me know about other topics that were going to be discussed in the sesion I can include them in the mail | 23:58 |
| n4nd0 | if that is ok for you of course | 23:59 |
| --- Log closed Fri Aug 03 00:00:02 2012 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!
