The crowding approach to niching in genetic algorithms
Evolutionary Computation
A dual-population genetic algorithm for adaptive diversity control
IEEE Transactions on Evolutionary Computation
A multiset genetic algorithm for the optimization of deceptive problems
Proceedings of the 15th annual conference on Genetic and evolutionary computation
Hi-index | 0.00 |
The traditional representation of populations used in evolutionary algorithms raises two types of problems: the loss of genetic diversity during the evolutionary process and evaluation of redundant individuals. To minimize these problems we developed MuGA, whose most distinctive feature is the representation of populations by multisets, and adapted the evolutionary process to handle the new representation. In this paper we present MuGA algorithm and explore its capacity to preserve the genetic diversity and find many optima using the Knapsack problem. Next we adapted genetic operators for the application of MuGA to real coded problems. The results obtained in a set of benchmark functions of these classes of problems, when compared with competitive algorithms, support our conviction that the multisets are an efficient representation of populations. Future work will focus on identifying limitations and subsequent improvements to the algorithm and its application to other classes of problems.