USING DISRUPTIVE SELECTION TO MAINTAIN DIVERSITY IN GENETIC ALGORITHMS

Genetic algorithms are a class of adaptive search techniques based on the principles of population genetics. The metaphor underlying genetic algorithms is that of natural evolution. With their great robustness, genetic algorithms have proven to be a promising technique for many o ptimization, design, control, and machine learning applications. A nov el selection method, disruptive selection, has been proposed. This met hod adopts a nonmonotonic fitness function that is quite different fro m conventional monotonic fitness functions. Unlike conventional select ion methods, this method favors both superior and inferior individuals . Since genetic algorithms allocate exponentially increasing numbers o f trials to the observed better parts of the search space, it is diffi cult to maintain diversity in genetic algorithms. We show that Disrupt ive Genetic Algorithms (DGAs) effectively alleviate this problem by fi rst demonstrating that DGAs can be used to solve a nonstationary searc h problem, where the goal is to track time-varying optima. Conventiona l Genetic Algorithms (CGAs) using proportional selection fare poorly o n nonstationary search problems because of their lack of population di versity after convergence. Experimental results show that DGAs immedia tely track the optimum after the change of environment. We then descri be a spike function that causes CGAs to miss the optimum. Experimental results show that DGAs outperform CGAs in resolving a spike function.