Multi-Objective Genetic Algorithm (MOGA)

Multi-objective Genetic Algorithm (MOGA) was proposed by Carlos and Peter which was inspired by the population genetics and the evolution of genes at the population level. In MOGA, the rank of an individual is relating to the number of chromosomes in the current population in which it is dominated. In this algorithm, all dominated MOGA individuals are assigned into Rank 1, while the dominated individuals are penalized based on the population density inside the corresponding trade-off surface. To assign the fitness, this algorithm first sort the population based on the rank. Then, the fitness is assigned to individuals by interpolating start from the Rank 1 to the worst rank (Rank ?). Finally, the fitness of individuals that sharing the same rank are averaged to ensure that all individuals will be sampled at the same rate. The fitness assignment procedures are executed to keep the global population fitness to be constant and maintaining appropriate selective pressure that is defined by the function used. However, this type of fitness assignment is likely may produce a large selection pressure that might cause premature convergence. Therefore, some improvements have been done by implementing a niche-formation method to allocate the population over the Pareto optimal region.