Open in new window / Try shogun cloud
--- Log opened Thu Feb 02 00:00:19 2012
shogun-buildbotbuild #309 of python_modular is complete: Failure [failed test_1]  Build details are at  blamelist: heiko.strathmann@gmail.com00:07
shogun-buildbotbuild #311 of python_modular is complete: Success [build successful]  Build details are at
shogun-buildbotbuild #307 of ruby_modular is complete: Failure [failed test_1]  Build details are at  blamelist:,, heiko.strathmann@gmail.com02:06
-!- dfrx [] has joined #shogun05:44
-!- ram108 [~ram@] has joined #shogun05:48
-!- ram108 [~ram@] has quit [Ping timeout: 240 seconds]06:12
-!- ram108 [~ram@] has joined #shogun06:42
-!- dfrx [] has quit [Remote host closed the connection]06:49
-!- dfrx [] has joined #shogun06:59
-!- ram108 [~ram@] has quit [Ping timeout: 240 seconds]07:52
-!- sonne|work [~sonnenbu@] has joined #shogun10:14
-!- dfrx [] has left #shogun []10:33
CIA-11shogun: Heiko Strathmann master * rede619a / examples/undocumented/python_modular/ : use MSE instead of accuracy and normal x-validation splitting -
CIA-11shogun: Soeren Sonnenburg master * r9ee39af / examples/undocumented/python_modular/ :10:46
CIA-11shogun: Merge pull request #362 from karlnapf/master10:46
CIA-11shogun: fix minor error in example -
CIA-11shogun: Heiko Strathmann master * r7083c78 / (3 files): renamed print to avoid python name clash -
CIA-11shogun: Soeren Sonnenburg master * rf0bdba7 / (3 files):10:59
CIA-11shogun: Merge pull request #363 from karlnapf/master10:59
CIA-11shogun: fixed warning -
CIA-11shogun: Heiko Strathmann master * rd9cb8d8 / src/shogun/evaluation/CrossValidation.cpp : added num runs check for confidence interval -
CIA-11shogun: Soeren Sonnenburg master * r3eab618 / src/shogun/evaluation/CrossValidation.cpp :11:15
CIA-11shogun: Merge pull request #364 from karlnapf/master11:15
CIA-11shogun: added check for confidence interval -
shogun-buildbotbuild #308 of ruby_modular is complete: Success [build successful]  Build details are at
-!- blackburn [~qdrgsm@] has joined #shogun13:11
blackburnsonne|work: the idea of any merging smells like a lot of pain in ass :)13:34
sonne|workblackburn: it is easy if someone just wants to contribute missing methods to shogun15:07
sonne|workotherwise very difficult agreed15:07
blackburndo you think he intended contributing to shogun?15:08
blackburnsonne|work: one idea I have is to suggest him to link his dimensionality reduction with EDRT15:08
sonne|workthe other alternative is he just wraps shogun objects in his tools15:12
blackburnsonne|work: how long does review by EIC editor take?15:12
blackburneditor in chief or so15:13
blackburnfirst stage in jmlr reviewing15:13
blackburnI'm just curious how much time does it take15:13
sonne|worklawrence is rather fast15:13
sonne|workbut the process will take between 3 and 6 months I would expect15:13
blackburncompletely crazy, why so much?15:14
blackburnthe only paper accepted in next month was gaussian processes toolkit15:14
blackburnsubmitted 8/1015:15
blackburnrevised 9/1015:15
blackburnpublished 11/1015:15
sonne|workit takes time to find action editors, it takes time to find reviewers, it takes time to get reviews, render decision, get the final version15:15
blackburnsonne|work: I see..15:17
blackburnsonne|work: in next 6 months I would implement more methods and paper would be outdated hah15:18
sonne|workthat's life15:20
blackburnsonne|work: oh, EIC (lawrence) is an author of locally linear embedding :)15:54
-!- Ram108 [~amma@] has joined #shogun16:05
blackburnRam108: hi16:06
Ram108i have almost finished the neural network-backpropagation algorithm16:07
Ram108i have a little more to make it complete though16:07
Ram108how do i make it shogun-like?16:07
blackburnwell just take a look on any classifier16:08
blackburnat least it should have methods train and apply :)16:08
Ram108make it similar to any of those right?16:08
Ram108lol yeah it has :)16:08
Ram108hmmm i ll modify the algorithm as per the requirement and email it to u :)16:08
Ram108would that be fine?16:09
blackburnwell pull request using github is preferrable16:09
Ram108oh okay i ll do that :)16:09
Ram108thanks :)16:09
Ram108have a great day :)16:10
Ram108by the way what else can i do?16:10
Ram108this project was really fun16:10
Ram108do u have any problems at hand i could read upon and implement?16:10
blackburnnot really16:11
Ram108hmmm thanks anyway :)16:11
blackburnbut take a look on github issues for shogun16:11
blackburnthere are some ideas16:11
Ram108oh cool :) i ll do that :)16:11
blackburnin fact they are a little complex16:12
blackburnbut at first we would have to polish your code16:12
Ram108i think am getting the hang of it....... lol i ll do that :)16:12
Ram108by the way what is generally required for me to get into GSOC 2012? i mean is there any pre-requisite?16:14
Ram108i ll email u the code before i use github16:14
blackburnyou would have to contact your possible mentor16:15
Ram108oh hmmm well mr soren?16:15
Ram108lol? but do u think its a little too late for that?16:15
Ram108i guess u r busy......... i ll contact u later then......... :)16:17
Ram108thanks for ur time :)16:17
Ram108thats okay lol i understand :)16:17
blackburnhang on :)16:17
Ram108sure :)16:17
blackburnit is not too late to contact mentors16:23
blackburngsoc wasn't announced yet16:23
blackburnso we didn't even create list of possible ideas16:23
Ram108oh hmmm16:24
Ram108by the way hw r u doin?16:24
blackburnfine, and you? ;)16:24
Ram108am doin great jusy busy thats all lol :)16:25
-!- n4nd0 [] has joined #shogun19:32
blackburnsonney2k: around?19:46
-!- Ram108 [~amma@] has quit [Ping timeout: 252 seconds]19:53
n4nd0I am having some trouble with the command addpath in octave20:15
n4nd0it doesn't seem to make the same job as it does in matlab20:15
n4nd0has anyone seen that before?20:15
blackburnunfortunately no20:15
n4nd0it looks like it adds the directory I am giving it correctly to the path, I have checked it with the command path20:16
n4nd0but it doesn't find the files that are there20:16
n4nd0let's see what I can do20:16
blackburnn4nd0: why do you need it?20:18
-!- Ram108 [~amma@] has joined #shogun20:18
n4nd0blackburn, it is for the directory to the training images20:18
n4nd0blackburn, instead of copying the one of my previous lab again to the working directory I am using now, I wanted just to use the one there20:19
blackburnI see20:19
n4nd0blackburn, mmm I don't know if I explained me clearly20:19
blackburnbtw yesterday I told you to create features.m too20:19
blackburnyou could avoid it20:19
blackburnjust create some feature matrix and store20:20
-!- Ram108 [~amma@] has quit [Client Quit]20:20
blackburnthen you could do only classifica20:20
blackburnin some script20:20
n4nd0I have already ported almost all the functions that I use to get the features from the images so it is not a problem ;)20:20
blackburnsounds promising :)20:21
blackburnn4nd0: for the moment I could tell you that possible gsoc ideas would include gaussian processes and structured output learning20:21
n4nd0blackburn, good, thanks :)20:22
n4nd0blackburn, I don't know about structured output learning but I could take it a look20:22
n4nd0blackburn, I have more experience with Gaussian processes, they appear everywhere!!20:23
blackburnwell then you could submit your proposal about GPs20:23
n4nd0we'll see20:23
n4nd0are you applying as well?20:24
n4nd0what is your project?20:24
n4nd0dimensionality reduction?20:24
blackburnI don't know yet20:24
n4nd0let's see if I am soon done with this part and I start working with shogun in python to try out different classifiers20:26
-!- n4nd0 [] has quit [Ping timeout: 256 seconds]21:06
CIA-11shogun: Soeren Sonnenburg master * r3a9377d / (4 files in 2 dirs):21:27
CIA-11shogun: Merge pull request #365 from karlnapf/master21:27
CIA-11shogun: added new error measure for regression (+5 more commits...) -
shogun-buildbotbuild #123 of nightly_default is complete: Success [build successful]  Build details are at
-!- n4nd0 [] has joined #shogun22:41
--- Log closed Fri Feb 03 00:00:19 2012