| --- Log opened Sun Oct 09 00:00:24 2011 | ||
| blackburn | yeah sonney2k v_array appears few times, ~100 errors :D | 00:11 |
|---|---|---|
| -!- alesis-novik [~alesis@cnat206.wlan.net.ed.ac.uk] has quit [Ping timeout: 258 seconds] | 00:33 | |
| -!- alesis-novik [~alesis@5ac57106.bb.sky.com] has joined #shogun | 01:08 | |
| -!- blackburn [~blackburn@85.114.187.69] has quit [Quit: Leaving.] | 02:10 | |
| -!- alesis-novik [~alesis@5ac57106.bb.sky.com] has quit [Ping timeout: 276 seconds] | 10:49 | |
| -!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 11:48 | |
| -!- blackburn [~blackburn@85.114.187.69] has joined #shogun | 11:55 | |
| -!- av3ngr [96cbdf9c@gateway/web/freenode/ip.150.203.223.156] has joined #shogun | 13:30 | |
| blackburn | sonney2k: around? | 14:23 |
| blackburn | I decided to not integrate covertree this time - JL's implementation is shity while DNCrane stl'y implementation is even slower than my | 15:46 |
| CIA-3 | shogun: Sergey Lisitsyn * r79758f3 / (3 files): Added use_arpack option for LLE, BLASed it a little more - http://git.io/vJEPTw | 16:34 |
| CIA-3 | shogun: Sergey Lisitsyn * r85a967c / src/shogun/preprocessor/KernelLocallyLinearEmbedding.cpp : BLASed KLLE a little - http://git.io/LUbZkw | 16:34 |
| -!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Quit: Leaving] | 17:56 | |
| -!- blackburn [~blackburn@85.114.187.69] has quit [Read error: No route to host] | 18:11 | |
| -!- blackburn [~blackburn@85.114.187.69] has joined #shogun | 18:12 | |
| -!- alesis-novik [~alesis@5ac57106.bb.sky.com] has joined #shogun | 19:16 | |
| blackburn | sonney2k: I miss you ;( | 19:38 |
| blackburn | :D | 19:38 |
| @sonney2k | blackburn, why did you give up? | 20:14 |
| blackburn | sonney2k: with covertree? | 20:14 |
| @sonney2k | yes | 20:14 |
| blackburn | JL's one crashes all the time | 20:15 |
| blackburn | DNCrane's one is even slower than it is now | 20:15 |
| blackburn | I have no time now to rewrite it | 20:15 |
| blackburn | and well it is not a bottleneck of my algos | 20:15 |
| blackburn | sonney2k: how about making my algos Machines? | 20:16 |
| @sonney2k | blackburn, does JL's crash when you apply it w/o your modifications? | 20:17 |
| blackburn | sonney2k: I just made it compilable, checked diff | 20:18 |
| blackburn | it crashes not often | 20:18 |
| blackburn | but I get wrong neighbors | 20:18 |
| blackburn | I spent two days trying to make it work - no success | 20:19 |
| blackburn | sonney2k: http://dl.dropbox.com/u/10139213/shogun/drawing-1.png how about that? | 20:23 |
| blackburn | I was lazy to make it pure uml but | 20:23 |
| @sonney2k | blackburn, if it crashes then you should create a reproducible test case and send it to JL | 20:39 |
| @sonney2k | or give me the test case and I will ask him | 20:39 |
| @sonney2k | blackburn, I think you can drop the shogun:: in these figures | 20:41 |
| @sonney2k | blackburn, what do you gain when making them machines? | 20:41 |
| blackburn | sonney2k: possible easier integration with heiko's parameter searching stuff, avoid things like I do in kpca and kernel lle | 20:44 |
| blackburn | but I see some impossible troubles: apply returns labels - I have no labels | 20:44 |
| blackburn | it could be still available as preprocessor by DimensionReductionPreprocessor proxy | 20:45 |
| -!- alesis-novik [~alesis@5ac57106.bb.sky.com] has quit [Quit: Leaving] | 22:42 | |
| --- Log closed Mon Oct 10 00:00:28 2011 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!
