Many monitoring, forecasting, and control operations occur in settings where relationships among key measurements must be learned quickly. Examples are on-line industrial processes where influent material is not consistent over time, energy load or price forecasting where demand characteristics change rapidly,and health management where relationships among monitored variables must be learned for each patient-treatment combination. The solution presented is a new neuro-computing system that learns in real-time, even when data arrival rates are several million measurements per second. The book describes benefits and features of the system, statistical foundations for the system, and several related models.
The book also describes available system software.
Presenting one practical solution to information processing problems, Jannarone (President, Rapid Clip Neural Systems, Inc.; Atlanta, GA) introduces his Rapid Learner software for parallel learning applications (e.g. industrial process monitoring and forecasting). He compares its features, foundations, operational details, and benefits to conventional programming, statistical, and neuro-computing methods. The text is not intended as a comprehensive survey of neural networks, machine learning models, learning theory, statistical or theoretical models, although it is directed at information scientists in these related fields. Annotation c. by Book News, Inc., Portland, Or.
