Design of efficient objective function for stochastic search algorithm

Abstract

In this paper, an approach to design the objective function for stochastic search algorithm is presented. N-Queen problem is solved employing genetic algorithm. Fitness function is one of the most critical parts of a genetic algorithm and its purpose is for parent selection and a measure for convergence. Problems with fitness are premature convergence and slow finishing. In general, a chromosome's value is computed by the order of its genes; any change in the order results in different chromosome's value. Here, a weight is computed to each bit (gene) position to compute the fitness of the string. To improve system performance, a weakest bit in the string is selected for cross over. It reduces the probability of dummy iterations and generation of recessive strings. Experimental results are compared with simple genetic algorithm, and enhanced improved genetic algorithm and adaptive genetic algorithm. Potential application includes search techniques and machine learning. © 2012 Springer-Verlag.