A generic jumping-gene paradigm

: concept, verification and applications

Abstract

A new evolutionary computing algorithm derived from the “Jumping genes” phenomenon is proposed in this thesis. It emulated the gene transposition in the genome that was firstly discovered by Nobel laureate, Barbara McClintock, from her work on the corn plant. The principle of jumping genes that was adopted for evolutionary computing is the main core of contribution in this thesis. It starts with the introduction of biological concept of jumping-genes and its required procedures for executing the computational multiobjective optimization are presented in details. A large number of unconstrained and constrained test functions were used for the verification of this new scheme and its performances on convergence and diversity were assessed. All the results were statistically examined and compared with those of other multiobjective evolutionary algorithms. It was found that this new scheme is robust and provides outcomes in speed with accuracy. Furthermore, from the obtained results, it was indicated that the jumping-gene paradigm is indeed a very competitive scheme for multiobjective optimization. Most importantly, it was shown that its superiority is the capability of providing widespread non-dominated solutions, especially those solutions at both extremes of the true Pareto-optimal or reference front. This new scheme was further applied to solve three typical real-world engineering problems. Again, the acquired results were all favorable and showed that it has the leading edge over other schemes when coming to the area of multiobjective optimization.

Date of Award	14 Jul 2006
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Kim Fung MAN (Supervisor)