Performance Analyses on Population Seeding Techniques for Genetic Algorithms

In Genetic Algorithm (GA), the fitness or quality of individual solutions in the initial population plays a significant part in determining the final optimal solution. The traditional GA with random population seeding technique is simple and proficient however the generated population may contain poor fitness individuals, which take long time to converge to the optimal solution. On the other hand, the hybrid population seeding techniques, which have the benefits of generating good fitness individuals and fast convergence to the optimal solution. Researchers have proposed several population seeding techniques using the background knowledge on the problem taken to solve. In this paper, we analyse the performance of different population seeding techniques for the permutation coded genetic algorithm based on the quality of the individuals generated. Experiments are carried out using the famous Travelling Salesman Problem (TSP) benchmark instances obtained from the TSPLIB, which is the standard library for TSP problems. The experimental results show the order of performance of different population seeding techniques in terms of Convergence Rate (%) and Error Rate (%). Keyword Genetic Algorithm, Population Seeding Technique, Travelling Salesman Problem, Performance Analysis, MATLAB