Further results on multiobjective evolutionary search for one-dimensional, density classifier, cellular automata, and strategy analysis of the rules

Artigo de evento
Data de publicação
Frontiers in Artificial Intelligence and Applications
Citações (Scopus)
Oliveira G.M.B.
Bortot J.C.
De Oliveira P.P.B.
Título da Revista
ISSN da Revista
Título de Volume
Membros da banca
A strong motivation for studying cellular automata (CA) is their ability to perform computations. However, the understanding of how these computations are carried out is still extremely vague, which makes the inverse problem of automatically designing CA rules with a predefined computational ability a fledgeling engineering endeavour. Various studies have been undertaken on methods to make CA design possible, one of them being the use of evolutionary computational techniques for searching the space of possible CA rules. A widely studied CA task is the Density Classification Task (DCT). For this and other tasks, it has recently been shown that the use of a heuristic guided by parameters that estimate the dynamic behaviour of CA can improve a standard evolutionary design. Considering the successful application of evolutionary multiobjective optimisation to several kinds of inverse problems, here one such technique known as Non-Dominated Sorting Genetic Algorithm is combined with the parameter-based heuristic, in the design of DCT rules. This is carried out in various alternative ways, yielding evolutionary searches with various numbers of objectives, of distinct qualities. With this exploration, it is shown that the resulting design scheme can effectively improve the search efficacy and obtain rules that solve the DCT with sophisticated strategies.
Assuntos Scopus
Automata rules , Cellular automatons , Density classification task , Dynamic behaviors , Evolutionary multiobjective optimization , Evolutionary search , Inverse designs , Non-dominated sorting genetic algorithm , Non-dominated sorting genetic algorithms , Parameter-based forecast of dynamic behavior
DOI (Texto completo)