75 search results - page 3 / 15 » Lagrangian Relaxation for MAP Estimation in Graphical Models |
ICML
2010
IEEE
13 years 6 months ago
2010
IEEE
Approximate MAP inference in graphical models is an important and challenging problem for many domains including computer vision, computational biology and natural language unders...
TIT
2008
13 years 4 months ago
2008
Abstract--Discrete-input two-dimensional (2-D) Gaussian channels with memory represent an important class of systems, which appears extensively in communications and storage. In sp...
DCC
2003
IEEE
13 years 11 months ago
2003
IEEE
- Rate-distortion theory is applied to the problem of joint compression and classification. A Lagrangian distortion measure is used to consider both the squared Euclidean error in ...
NIPS
2007
13 years 7 months ago
2007
We present a novel message passing algorithm for approximating the MAP problem in graphical models. The algorithm is similar in structure to max-product but unlike max-product it ...
CVPR
2004
IEEE
14 years 7 months ago
2004
IEEE
This paper proposes a new framework for image segmentation based on the integration of MRFs and deformable models using graphical models. We first construct a graphical model to r...