Watercooler: 29 April 2005
The Scientific Watercooler meetings, hosted by the Center for Computation and Visualization, are informal gatherings of researchers who need to use high performance computing to enable their research. Each month, these meetings are held at 4:00PM in the Swig Boardroom, 2nd floor, CIT. This month's agenda will include the following topics:
Learning the Spatial Statistics of Images: Models and Computational Challenges
Presented by Stefan Roth, Graduate Student, Computer Science
This talk will introduce the problem of modeling the spatial statistics of images along with a few examples from an image reconstruction application. It will then discuss how we learn a new higher-order Markov random field model that can capture the rich spatial structure of images. The talk will conclude with a discussion of the computational problems that we had in training this model, and how we solved them using a combination of MATLAB and MPI.
CCV Staff
Open Discussion
Light refreshments will be served.