Investigation of the Theoretical Foundations of Self-Organizing and Learning Systems.

摘要

Techniques for terminal-value control of vehicles and for control systems that must avoid operating regions that involve high performance penalties were studied. Search algorithms that possess long-term memory were formulated to cope with both cases. Because the processes used for gaining information during control,retaining it,and accessing this information determine,in large measure,the capabilities of self-organizing and learning systems,theoretical factors pertaining to realization of efficient memory processes for these systems received special consideration. (Modified author abstract)