|Stochastic Systems Group|
SSG Doctoral Student
In this talk, we discuss the class of multiscale models, which is well-suited to represent a wide variety of random processes and in addition, admits an extremely efficient estimation algorithm. These types of models have proven useful in a number of applications including image processing, remote sensing, and geophysics.
We focus on the problem of constructing and updating multiscale models given data of the process of interest. Previous results provide a non-iterative algorithm for realizing a multiscale model given complete covariance information. However, for problems of even moderate size, knowledge of the full covariance is impractical. For this reason, we seek methods to construct multiscale models based solely on sample paths of the process, with no assumed knowledge of the covariance. In addition, we show how these types of models may be efficiently updated given a new set of data.
Problems with this site should be emailed to firstname.lastname@example.org