Building High-level Features Using Large Scale Unsupervised Learning
3. ILSVRC 2012 大è¦æ¨¡ç‰©ä½“èªè˜ã®ã‚³ãƒ³ãƒ†ã‚¹ãƒˆ http://www.image-net.org/challenges/LSVRC/2012/ Classification Localization Team name Error Team name Error 1 Super Vision 0.15315 1 Super Vision 0.335463 2 Super Vision Deep LearningVision 0.16422 2 Super 0.341905 3 ISI 0.26172 3 OXFORD_VGG 0.500342 4 ISI 0.26602 4 OXFORD_VGG 0.50139 5 ISI 0.26646 5 OXFORD_VGG 0.522189 6 ISI 0.26952 6 OXFORD_VGG 0.529482 7 OXFORD_VGG
Large Scale Learning to Rankã‚’èªã‚“ã - å°„æ’ƒã—ã¤ã¤å‰è»¢ 改
本当ã¯ä¸‰ãŒæ—¥ä¸ã«ã¾ã¨ã‚‚ãªã‚¨ãƒ³ãƒˆãƒªã‚’1本ãらã„ã¯æ›¸ã予定ã ã£ãŸã®ã ãŒã€ã¡ã‚‡ã£ã¨ç„¡ç†ã ã£ãŸã€‚ã¨ã„ã†ã‚ã‘ã§ã€å®Ÿè³ªçš„ã«æ–°å¹´ä¸€æœ¬ç›®ã®ã‚¨ãƒ³ãƒˆãƒªã§ã™ã€‚Large Scale Learning to Rank (D. Sculley, NIPS Workshop on Advances in Ranking, 2009) (pdf) ã‚’èªã‚“ã ã®ã§ã€1本目ã®ã‚¨ãƒ³ãƒˆãƒªã¨ã—ã¦ã“ã®è«–文を紹介ã—ãŸã„。 ã§ã¯æ—©é€Ÿæœ¬é¡Œã«å…¥ã‚ã†ã€‚é †ä½å¦ç¿’ã«ãŠã„ã¦ã€Pairwise Learningã‚’å˜ç´”ã«è¡Œã†ã¨ã€n^2ã®å¦ç¿’コストãŒã‹ã‹ã‚‹ã€‚ã“ã‚Œã¯è¨ˆç®—時間ã¨ã—ã¦ã¯åŽ³ã—ã„部類ã«å…¥ã‚‹ã€‚ãã‚‚ãã‚‚é †ä½å¦ç¿’ã£ã¦ãªã«ã€ã¨ã„ã†äººã¯ã€WWW2009ã®ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«(pdf)ã¨ã‹ã‚’å‚ç…§ã—ã¦ãã ã•ã„。 Bottouらã¯ã€SGDã®ä¸€èˆ¬åŒ–能力ã¯ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆã®ã‚µã‚¤ã‚ºã«ä¾ã‚‰ãšã€ã©ã‚Œã ã‘ã®stochastic stepを実行ã—ãŸã‹ã§æ±ºã¾ã‚‹ã¨è¨€ã†äº‹ã‚’示ã—ãŸã€‚ãã“ã§ã€Sc
lwlm, The Latent Words Language Model.
Daichi Mochihashi NTT Communication Science Laboratories $Id: lwlm.html,v 1.1 2010/03/19 10:15:06 daichi Exp $ lwlm is an exact, full Bayesian implementation of the Latent Words Language Model (Deschacht and Moens, 2009). It automatically learns synonymous words to infer context-dependent "latent word" for each word appearance, in a completely unsupervised fashion. Technically, LWLM is a higher-or
1
リリースã€éšœå®³æƒ…å ±ãªã©ã®ã‚µãƒ¼ãƒ“スã®ãŠçŸ¥ã‚‰ã›
最新ã®äººæ°—エントリーã®é…ä¿¡
処ç†ã‚’実行ä¸ã§ã™
j次ã®ãƒ–ックマーク
kå‰ã®ãƒ–ックマーク
lã‚ã¨ã§èªã‚€
eコメント一覧を開ã
oページを開ã
{{#tags}}- {{label}}
{{/tags}}