Directed mutation in genetic algorithms

Abstract

Mutation is an important genetic operation that helps to maintain the genetic diversity of the population in order to achieve a good solution to an optimization problem. The conventional mutation, with its random nature and low probability of occurrence, is unable to guide the algorithm for fast convergence. In this article, we propose a new mutation technique called, directed mutation following the concept of induced mutation in biological systems. Directed mutation deterministically introduces new points in the population guided (directed) by the information acquired in the previous generations. It expedites the algorithm using an acceleration function which can be based on gradient or extrapolation. The effectiveness of the new technique has been demonstrated through optimization of a set of complex functions, and the results have also been compared with those of the conventional GAs.