Machines that Learn: Based on the Principles of Empirical Control

This important new book details the design of robotic devices that can learn--without input from outside sources--from their own experience and are thus able to establish self-sustaining behavior. The subject of considerable growing interest across a wide range of fields, "learning machines" will provide invaluable assistance in both the home and the workplace; some functions include monitoring industrial equipment, piloting aircraft, and performing many routine jobs without special instructions. Machines that Learn shows the design of a simple neural network that works more like the animal brain than conventional neural networks. It also provides unusual explanations of the terms "organization", "information", and "order". Hundreds of drawings and diagrams provide invaluable supplements to the text. Uniquely accessible and well-written, the book will be welcomed by researchers, technicians, engineers, and students interested in artificial intelligence, control engineering, computer science, and robotics.