Visualization of evolutionary process in genetic programming
Chatchawan Wongsiriprasert, Prabhas Chongstitvatana, Somchai Prasitjutrakul · 1998
Genetic programming (GP) is a process of finding solutions of a problem through the evolution of population of computer programs to solve that problem. As the pattern of searching for solutions in a problem landscape can be quite complicate, it is difficult to explain, in many cases, some phenomena that happened during the evolutionary process. This work describes an attempt to visualize that evolutionary process based on a case study of using GP to solve a problem in robot learning. There are many parameters in the system that affected its performance. We are especially interested in how the genetic operations affected individuals in the population. The result from "understanding" of the evolutionary process via visualization has been used to explain some phenomena in our GP experiments. 1. Introduction Visualization of algorithm and data is a means to study how algorithms work by using graphical views and animations of the algorithms and data in action [3, 4]. This work describes a...