An Improved Clustering Based Genetic Algorithm for Solving Complex NP Problems

Abstract

Problem statement: The selection process is a major factor in genetic algorithm which determines the optimality of solution for a complex problem. The selection pressure is the critical step which finds out the best individuals in the entire population for further genetic operators. The proposed algorithm tries to find out the best individuals with reduced selection pressure than standard genetic algorithm which is commonly used. Approach: The selection process is refined in the proposed algorithm by using the concept of clustering rather than traditional selection mechanisms like Roulette wheel selection, Rank selection, Tournament selection. Results: As the selection process is improved in our approach, the convergence velocity of the genetic algorithm is improved by reaching the optimal solution quickly and the optimality of the solution is also fine-tuned. Conclusion: A new variant of the standard genetic algorithm is proposed which reduces the execution time of the algorithm by gearing up the selection process to reach the most efficient solution. The fit individuals are selected for crossover and mutation in all generations thereby reaching the solution without much complex process.