Improving Strategy Parameters of Evolutionary Computations with Interactive Coordinated Views

Evolutionary algorithms (EAs) use mechanisms inspired by biological evolution, e.g., natural selection, recombination, or mutation, that work on populations of solutions for a specific problem. These recurring processes produce a huge amount of time-varying data. In order to get a better insight into the progress of EAs, a Java-based visualization tool, called EAVis, was developed. The most important aims of EAVis are the selection, concentration, and abstraction of evolutionary data on different levels using a variety of visualization methods. Several coordinated views (2D and 3D) help the user to watch each generation step of the EA and to derive knowledge as well as better understanding of the underlying evolutionary computational models. Among other things, this is also important for a clever parameter setting to gain better performance values.