Structural optimization by genetic algorithm with degeneration (GA d )

Genetic Algorithms (GAs) are models of biological evolution. But the concept of degeneration had not previously been introduced to GA. Degeneration is the evolutionary decline or loss of a function or characteristic, and it had not been thought that degeneration could be used for optimization. In this paper, GA with degeneration (GAd), which introduces the concept of degeneration to GA, is proposed. It is assumed that damaged genes and irreversible mutations cause degeneration and that if degeneration occurs, some characteristics of an individual are lost. If the characteristics are regarded as parameters in a model, the number of parameters can be reduced and the optimal parameter structure can be discovered. It is shown that the optimal parameter structure of some models such as neural networks can be discovered by GAd. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(5): 32–43, 2004; Published online in Wiley InterScience (). DOI 10.1002sscj.10552