Wang C., Hill D.J. Deterministic Learning Theory for Identification, Recognition and Control

CRC Press, 2010. — 218 p.The problem of learning in dynamic environments is important and challenging. In the 1960s, learning from control of dynamical systems was studied extensively. At that time, learning was similar in meaning to other terms such as adaptation and self-organizing. Since the 1970s, learning theory has become a research discipline in the context of machine learning, and more recently as computational or statistical learning. As a result, learning is considered as a problem of function estimation on the basis of empirical data, and learning theory has been studied mainly by using statistical principles. Although many problems in learning static nonlinear mappings have been handled successfully via statistical learning, a learning theory for dynamic systems, for example, learning of the functional system dynamics from a dynamical process, has received much less investigation.
This book emphasizes learning in uncertain dynamic environments, in which many aspects remain largely unexplored. The main subject of the monograph is knowledge acquisition, representation, and utilization in unknown dynamic processes. A deterministic framework is regarded as suitable for the intended purposes. Furthermore, this view comes naturally from deterministic algorithms in identification and adaptive control of nonlinear systems which motivate some of our work. Referred to as deterministic learning (DL), the learning theory presented gives promise of systematic design approaches for nonlinear system identification, dynamic pattern recognition, and intelligent control of nonlinear systems.RBF Network Approximation and Persistence of Excitation
The Deterministic Learning Mechanism
Deterministic Learning from Closed-Loop Control
Dynamical Pattern Recognition
Pattern-Based Intelligent Control
Deterministic Learning with Output Measurements
Toward Human-Like Learning and Control