Applying visualizing techniques to evolutionary computation (EC) enables users to see the search behavior of their algorithms. Consequently, they can attribute credit to the individual designs and judge the quality of each algorithm based on its ability to explore the problem space. In this book chapter, Collins did a wonderful job by reviewing the prior work in the area of applying visualization to explore the capability of genetic algorithms (GAs) for solving optimization problems (for instance, the traveling salesman problem).
Specifically, this chapter starts with the introduction of techniques for visualizing the quality of solutions, such as 2D fitness graphs, 3D fitness graphs, and alternative plots. It then describes techniques for producing problem-specific visualization of an evolutionary algorithm’s individual parts. Efforts for avoiding overwhelming users with too much information are also depicted. Techniques for navigating the algorithm’s execution follow, such as the one that could provide single-step controls. Other techniques, such as how to edit the algorithm’s parameter during the course of executing the algorithm, were also given in this chapter.
Undoubtedly, this chapter is a good literature review for researchers who are interested in this area. However, while I was reading the chapter, I kept wondering how much users have benefited from these techniques. Did anyone discover or make any improvement on the algorithm by using these techniques?