Designing Evolutionary Algorithms for Dynamic Environments

The robust capability of evolutionary algorithms (EAs) to find solutions to difficult problems has permitted them to become popular as optimization and search techniques for many industries. Despite the success of EAs, the resultant solutions are often fragile and prone to failure when the problem changes, usually requiring human intervention to keep the EA on track. Since many optimization problems in engineering, finance, and information technology require systems that can adapt to changes over time, it is desirable that EAs be able to respond to changes in the environment on their own. This book provides an analysis of what an EA needs to do to automatically and continuously solve dynamic problems, focusing on detecting changes in the problem environment and responding to those changes. In this book we identify and quantify a key attribute needed to improve the detection and response performance of EAs in dynamic environments. We then create an enhanced EA, designed explicitly to exploit this new understanding. This enhanced EA is shown to have superior performance on some types of problems. Our experiments evaluating this enhanced EA indicate some pre­ viously unknown relationships between performance and diversity that may lead to general methods for improving EAs in dynamic environments. Along the way, several other important design issues are addressed involving com­ putational efficiency, performance measurement, and the testing of EAs in dynamic environments.

1 Introduction.- 2 Problem Analysis.- 3 Solutions from Nature and Engineering.- 4 Diversity Measurement.- 5 A New EA for Dynamic Problems.- 6 Experimental Methods.- 7 Performance Measurement.- 8 Analysis and Interpretation of Experimental Results.- 9 Experimental Results for Population Initialization.- 10 Summary and Conclusion.- Notation.- References.

Dr. Morrison has been at Mitretek Systems for four years as a Senior Manager and Fellow. He currently serves as an advisor to U.S. government officials regarding advanced software development projects. Previously, Dr. Morrison was Chief Scientist for the SWL division at GRC International, where he was responsible for product development and innovation involving new techniques and applications in the areas of data visualization, computational intelligence, machine learning, and high-speed decision support systems. His accomplishments at GRCI include the creation of a novel genetic-algorithm based decision-support system for commodity traders, development of a method for integrating quantitative and qualitative information for a U.S. government agency, and the framework design for a commercial software-based intelligent agent for use by the Defense Advanced Research Projects Agency. Before joining GRCI, Dr. Morrison was Director of Software Engineering at Hughes Training, Inc., developing high-fidelity, real-time flight simulators for U.S. and foreign military customers.

Dr. Morrison has presented multiple papers at major internatinal conferences on Evolutionary Compuation, has served as the Technical Director for the Software Program Manager's Network and is a past member of the Airlie Software Council. He was an invited speaker at the initial meeting of the Narional Software Alliance in 1998 and at the AIE-sponsored Annual Conference on Software Metrics. He holds a B.S. in Aeronautical and Astronautical Engineering from Purdue University, an M.B.A. from Southern Illinois University, and a Ph.D. in Information Technology from George Mason University.