| --- Log opened Sat May 07 00:00:47 2011 | ||
| -!- blackburn [~qdrgsm@188.122.225.203] has quit [Quit: Leaving.] | 01:22 | |
| -!- blackburn [~qdrgsm@188.122.225.203] has joined #shogun | 01:23 | |
| -!- serialhex [~quassel@99-101-148-183.lightspeed.wepbfl.sbcglobal.net] has quit [Read error: Connection reset by peer] | 01:55 | |
| -!- serialhex [~quassel@99-101-148-183.lightspeed.wepbfl.sbcglobal.net] has joined #shogun | 01:55 | |
| -!- blackburn [~qdrgsm@188.122.225.203] has quit [Quit: Leaving.] | 02:03 | |
| -!- blackburn [~qdrgsm@188.122.225.203] has joined #shogun | 11:41 | |
| -!- blackburn [~qdrgsm@188.122.225.203] has quit [Quit: Leaving.] | 15:10 | |
| -!- blackburn [~qdrgsm@188.122.225.203] has joined #shogun | 15:54 | |
| blackburn | sonney2k: around? | 18:54 |
|---|---|---|
| @sonney2k | blackburn, now | 21:09 |
| blackburn | sonney2k: how are you? | 21:09 |
| blackburn | I was confused with heiko's subsets but I think I won't use it | 21:09 |
| @sonney2k | blackburn, apart from tired no news | 21:09 |
| @sonney2k | blackburn, I don't understand? | 21:10 |
| @sonney2k | we use this for data splitting... | 21:10 |
| blackburn | sonney2k: in LLE I could use subsets | 21:10 |
| @sonney2k | blackburn, I guess you can use subsets internally within LLE though... | 21:11 |
| blackburn | ok | 21:12 |
| blackburn | sonney2k: btw, here will be matrices of greaat sizes, O(n*dim), where n is number of examples | 21:18 |
| blackburn | I don't like it exactly, but seems there is no other way | 21:19 |
| @sonney2k | blackburn, big your mean | 21:19 |
| blackburn | seems so | 21:19 |
| @sonney2k | then the algorithm is only suitable for toy data | 21:19 |
| @sonney2k | you could ask cwidmer / mlsec | 21:19 |
| blackburn | ah, even more | 21:20 |
| @sonney2k | maybe the know better algos - but a baseline is also worth having | 21:20 |
| blackburn | sonney2k: I think it is a good algo anyway | 21:20 |
| @sonney2k | then one just does dim reduction on a subset of the data | 21:20 |
| blackburn | ISOMAP is even slower | 21:21 |
| blackburn | sonney2k: btw how it can be used in bioinformatics? with mapping A->1, C->2, etc? | 21:23 |
| blackburn | or with kernels? | 21:23 |
| @sonney2k | blackburn, both - depends... | 21:23 |
| blackburn | unfortunately I don't know anything about this field :( | 21:23 |
| @sonney2k | A-> 0001 , C -> 0010 | 21:23 |
| @sonney2k | etc | 21:23 |
| blackburn | I see | 21:24 |
| @sonney2k | doesn't matter... | 21:24 |
| blackburn | so in this case LLE will produce huge matrices | 21:24 |
| blackburn | as I know there are >200K features, is it? | 21:24 |
| @sonney2k | blackburn, yeah useless in this scenario | 21:26 |
| blackburn | sonney2k: we discussed possibility of writing a paper about dimreduction with shogun to some journal | 21:26 |
| blackburn | sonney2k: he said it is possible with "Open Research Computation" | 21:27 |
| blackburn | but while I was reading FAQ, I became surprised a lot with "We charge a flat fee of £995/$1660/€1120 for each accepted manuscript." | 21:28 |
| blackburn | sonney2k: is it a common practice? | 21:28 |
| blackburn | I'm a bit newbie with these issues | 21:32 |
| @sonney2k | sorry got disconnected | 21:34 |
| blackburn | repeat? | 21:35 |
| blackburn | sonney2k: hoho! seems I have done 1/3 of LLE | 21:44 |
| -!- serialhex_ [~quassel@99-101-148-183.lightspeed.wepbfl.sbcglobal.net] has joined #shogun | 23:43 | |
| -!- serialhex [~quassel@99-101-148-183.lightspeed.wepbfl.sbcglobal.net] has quit [Read error: Connection reset by peer] | 23:43 | |
| -!- serialhex_ is now known as serialhex | 23:43 | |
| blackburn | serialhex: how are ya? | 23:44 |
| serialhex | good, how about yourself blackburn? | 23:44 |
| blackburn | nice, implemented an LLE a little | 23:45 |
| serialhex | 1/3 of one?? :P | 23:45 |
| @bettyboo | serialhex, funny | 23:45 |
| blackburn | serialhex: yeap, I consider so because it have three stages :) | 23:46 |
| serialhex | cool | 23:46 |
| blackburn | serialhex: is 'I'm finished' a proper english? | 23:46 |
| serialhex | yes, though "is 'I'm finished' proper english?" would be better :D | 23:48 |
| blackburn | :D | 23:48 |
| blackburn | I'm useless | 23:49 |
| serialhex | no, youre getting better! | 23:49 |
| --- Log closed Sun May 08 00:00:48 2011 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!