Behavioral analysis of genetic algorithm for function optimization

Abstract

Function optimization is the process of finding absolutely best values of the variables so that value of an objective function becomes optimal. Many optimization techniques are available but if they perform well on one class of problems then they may not work at all on other classes of problems. Moreover function optimization problems are a class of NP-complete problems so there is not a single algorithm that solves these problems in polynomial time. Genetic algorithm is probabilistic, heuristic, robust search algorithm premised on the evolutionary ideas of natural selection and genetic. Main idea behind the design of genetic algorithm is to achieve robustness and adaptiveness in real world complex problems. Genetic algorithm can be viewed as an optimization technique, which exploits random search within a defined search space to solve a problem, by some intelligence ideas of nature. Genetic algorithm can be used as a general function optimizer that can solve problems of any classes. Genetic algorithm is widely used in many applications but it has few weak points like selection pressure, balance between crossover and mutation, representation problems, stochastic nature of crossover etc. A new model of genetic algorithm called Guided model that is more deterministic, guided and uses less stochastic information. The proposed Guided model is designed such a way that it does not suffer from the problems of selection pressure, lost of best chromosomes, crossover rates, etc Results shows that proposed Guided model gives far better results compared to Holland model and commonly used Common model.