Sieves and Signal Extraction.
Abstract
Given a Gaussian signal with known mean and covariance kernel R, and given a second covariance R, the PI has proved several results concerning the probability that the sample path of the signal will fall in the reproducing kernal Hilbert space with kernal R. This is applied to optimal extraction of an unobservable signal based on its conditional mean, and to generalization of a zero-one law given by Kallianpur and by Driscoll. For an observable Gaussian process with unknown mean and covariance, an extension of previous work shows that simultaneous consistent estimation of both the mean and the covariance is possible by the method of sieves. In both this and the signal extraction problem, no assumption is made about the nature of the time parameter of the process. Some work on the axiomatic theory of confounding in experimental designs is also reported.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 29, 1986
- Accession Number
- ADA177628
Entities
People
- Jay H. Beder
Organizations
- University of Wisconsin–Milwaukee