1068 search results - page 22 / 214 » Extremely randomized trees |
JAIR
2008
13 years 8 months ago
2008
In this paper, we show that a continuous spectrum of randomisation exists, in which most existing tree randomisations are only operating around the two ends of the spectrum. That ...
ECCV
2010
Springer
13 years 8 months ago
2010
Springer
Abstract. Multiple-instance learning (MIL) allows for training classifiers from ambiguously labeled data. In computer vision, this learning paradigm has been recently used in many ...
SIAMDM
2010
13 years 3 months ago
2010
We prove that if T is a tree on n vertices with maximum degree and the edge probability p(n) satisfies: np C max{ log n, n } for some constant > 0, then with high probability...
FOCS
2009
IEEE
14 years 3 months ago
2009
IEEE
In this paper, we set forth a new algorithm for generating approximately uniformly random spanning trees in undirected graphs. We show how to sample from a distribution that is wi...
FOCS
1989
IEEE
14 years 21 days ago
1989
IEEE
We present a randomized strategy for maintaining balance in dynamically changing search trees that has optimal expected behavior. In particular, in the expected case a search or a...