Identifying an Unknown Static Process Having Uniformly Distributed Observational Errors,
Abstract
The problem of identifying a black box having uniformly distributed errors in the observations is investigated. In particular, those black boxes are studied for which the space of candidate models is infinite dimensional. Since such black boxes cannot be identified deterministically by using a finite amount of data, the concept of a probability of identification is followed. An adequacy criterion for modeling is defined in which a model is adequate provided it passes a finite sequence of tests, the length of which depends on the desired probability of identification. An algorithm is presented for determining an adequate model within the space of polynomial functions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 02, 1975
- Accession Number
- ADA008551
Entities
People
- Harold L. Stalford
Organizations
- United States Naval Research Laboratory