Processing of Prior-Information in Statistics by Projections on Convex Cones

Autor: E. Rödel
Rok vydání: 1997
Předmět:
Zdroj: Learning, Networks and Statistics ISBN: 9783211829103
DOI: 10.1007/978-3-7091-2668-4_4
Popis: We investigate a Random-Search-Algorithm for finding the projection on a closed convex cone in R P with respect to a norm defined by any positive definite matrix. It is shown that this algorithm converges almost surely. The power of the algorithm is demonstrated by examples from statistics in which processing of prior information may be formulated as projections of parametervectors on polyhedral cones.
Databáze: OpenAIRE