ABSTRACT
Genetic algorithms (GA) have emerged as effective search and optimization methods with applications in several problem domains. When the underlying search space has several locally optimal solutions apart from the globally optimal solution, GAs emerge as worthy alternatives to traditional optimization techniques. In this chapter, the authors discuss the tradeoffs between exploration and exploitation that a GA needs to make to optimize its search. They analyze how adaptive strategies for dynamically modifying the control parameters can lead to improved GA-search. The authors also discuss the motivating factors for employing adaptive control parameters. They describe an approach of using adaptively varying probabilities of crossover and mutation for multimodal function optimization. The authors presents experimental results to compare the performance of the GAs with and without adaptive probabilities of crossover and mutation. They conduct extensive experiments on a wide range of problems including traveling salesman problem, neural network weight-optimization problems, and generation of test vectors for VLSI circuits.
