A new evolutionary computing algorithm on the basis of the "jumping genes" (JG) phenomenon is proposed in this paper. It emulates the gene transposition in the genome that was discovered by Nobel Laureate, Barbara McClintock, in her work on the corn plants. The principle of JGs that is adopted for evolutionary computing is outlined. The procedures for executing the computational optimization are provided. A large number of constrained and unconstrained test functions have been utilized to verify this new scheme. Its performances on convergence and diversity have been statistically examined and comparisons with other evolutionary algorithms are carried out. It has been discovered that this new scheme is robust and able to provide outcomes quickly and accurately. A stringent measure of binary ε-indicator is also applied for algorithm classification. The outcome from this test indicates that the JG paradigm is a very competitive scheme for multiobjective optimization and also a compatible evolutionary computing scheme when speed in convergence, diversity, and accuracy are simultaneously required. © 2007 IEEE.
Bibliographical noteThis work was supported by City University of Hong Kong under Grant SRG 7002025.
- Genetic algorithms (GAs)
- Jumping genes (JGs)
- Multiobjective evolutionary algorithms (MOEAs)
- Test functions