70 search results - page 6 / 14 » Bypassing the embedding: algorithms for low dimensional metr... |
IJCV
2007
14 years 9 months ago
2007
The parameter space for the ellipses in a two dimensional image is a five dimensional manifold, where each point of the manifold corresponds to an ellipse in the image. The parame...
ICPR
2006
IEEE
15 years 10 months ago
2006
IEEE
We present a dynamic inference algorithm in a globally parameterized nonlinear manifold and demonstrate it on the problem of visual tracking. An appearance manifold is usually non...
114
click to vote
MMM
2011
Springer
14 years 1 months ago
2011
Springer
Abstract. Efficient nearest neighbor (NN) search techniques for highdimensional data are crucial to content-based image retrieval (CBIR). Traditional data structures (e.g., kd-tree...
NIPS
2004
14 years 11 months ago
2004
Embedding algorithms search for low dimensional structure in complex data, but most algorithms only handle objects of a single type for which pairwise distances are specified. Thi...
ICML
2008
IEEE
15 years 10 months ago
2008
IEEE
In this paper we perform an empirical evaluation of supervised learning on highdimensional data. We evaluate performance on three metrics: accuracy, AUC, and squared loss and stud...