In this section:
In this section, three frequently used fitness functions are described. Two of these fitness functions can be used to solve virtually any problem of symbolic regression. One of them is based on the absolute error and the other on the relative error. The third fitness function is a small variant of the already familiar fitness function used in the previous section to solve the majority function problem and is commonly used to solve most problems of logic synthesis. Other examples of more exotic fitness functions are given throughout the book. We will see that the success of a problem greatly depends on the way the fitness function is designed. Therefore, in the design of the fitness function, the goal must be clearly and correctly defined in order to make the system evolve in the intended direction.
|