Bayesian decision making and learning for continuous-time Markov systems
Erdal Panayırcı · Michigan State University Libraries · 1970
Abstract : The document is concerned with Bayesian decision making and learning algorithms for a particular problem in parametric pattern recognition in which each of a finite set of pattern classes is characterized by a continuous-time, discrete-state Markov process. The basic problem considered is that of determining rules for making decisions about the identity of the active pattern class based upon observation of a sample function in some finite interval. The stationary transition probability matrices for the processes in question are the parameters of the pattern classes. (Author)