11484 search results - page 55 / 2297 » Modeling in the Large and Modeling in the Small |
ECCV
2010
Springer
14 years 2 months ago
2010
Springer
Abstract. Most current approaches to recognition aim to be scaleinvariant. However, the cues available for recognizing a 300 pixel tall object are qualitatively different from tho...
IJBC
2006
13 years 8 months ago
2006
IJBC
2006
13 years 8 months ago
2006
NAACL
2004
13 years 10 months ago
2004
Previous work demonstrated that web counts can be used to approximate bigram frequencies, and thus should be useful for a wide variety of NLP tasks. So far, only two generation ta...
HAIS
2010
Springer
14 years 1 months ago
2010
Springer
The intuition behind ensembles is that different prediciton models compensate each other’s errors if one combines them in an appropriate way. In case of large ensembles a lot of...