Component-based visual evolutionary computation

Component-based computation advocates for building systems by assembling prefabricated self-contained software modules within a formally specified composition model. Furthermore, visual components allow users to perform such assemblage in a friendly graphical board were components are connected visually with virtually no coding needed. Recently, an Evolutionary Computation (EC) suite of components called Goldenberry complying with such premises has been released. In this paper we demonstrate its feasibility in several EC tasks: optimisation of continuous-valued and discrete-valued costs functions using Genetic Algorithms and Estimation of Distribution Algorithms as well as feature selection algorithms and relevance estimation in data mining tasks. We discuss the advantages and weaknesses of the approach and the toolbox, its differences with other existing software tools and proposing avenues of future development. Goldenberry is open-source under the New BSD License and it is available at.