NONSUPERVISED PATTERN RECOGNITION THROUGH THE DECOMPOSITION OF PROBABILITY FUNCTIONS.
Abstract
Two problems of parametric statistics are investigated with a view to their application to nonsupervised pattern recognition. Each of the problems can be described as follows: given a random sample drawn from a finite mixture of probability functions, where each element of the mixture is of a known parametric form, determine the unknown parameters of the mixture, f(X). The problem is treated in two parts. In the first part, it is assumed that the function f(X) is known and the decomposition of f(X) into its components is discussed. The second part deals with the estimation of f(X) on the basis of a random sample drawn according to it. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1966
- Accession Number
- AD0637486
Entities
People
- Donald F. Stanat
Organizations
- University of Michigan