Practical Neural Network Recipies in C++

154 Chapter 8

The first step in the genetic subroutine is to build the initial

population. We keep track of the best individual at all times, both now

and later. We may need this individual if the climb parameter is non-

zero.

More importantly, it is this individual, not necessarily the best

in the current generation, which will be returned to the user when

operation is complete. While the initial population is being built, we

only need to note the index of the best, rather than its actual parame-

ter values. After the population is complete, the best member (pointed

to by this index) will be preserved in the best work area.

Two population wor ...