Practical Neural Network Recipies in C++

Eluding Local

Minima II

155

Some Functions for Genetic Optimization

This section will present three similar functions for the genetic

optimization routine to minimize. Although they differ in only small

ways,

their behavior under genetic optimization is significantly

different. These differences are especially great if the crossover

probability is high and the population size is moderate (500-1000). It

behooves the interested reader to study these fimctions carefully and

understand what is responsible for their varying behavior.

In the section on designing the chromosome's structure, it was

pointed out that genes which interact should be place ...