Publication Details
Evolution of Cellular Automata Development Using Various Representations
cellular automaton; transition function; development; evolutionary algorithm
This paper introduces a comparative summary regarding evolution of multistate cellular automata by means of various representations of transition functions. In particular, a conventional table-based representation and an advanced approach using Conditionally Matching Rules is applied. The French flag development from a seed is considered as a case study task. The results show some remarkable differences in the cellular automata behaviour which is evidently caused by the representation used. This includes the issue of emergence of the pattern from chaotic state or rather its systematic construction, stability of the pattern and limitation of its successful development to fixed-size automata only. The comparison of these results is enabled by using a custom variant of genetic algorithm that provides working solutions of this task using both representations of the transition function.
@INPROCEEDINGS{FITPUB11902, author = "Michal Bidlo", title = "Evolution of Cellular Automata Development Using Various Representations", pages = "107--108", booktitle = "GECCO '19 Proceedings of the Genetic and Evolutionary Computation Conference Companion", year = 2019, location = "Praha, CZ", publisher = "Association for Computing Machinery", ISBN = "978-1-4503-6748-6", doi = "10.1145/3319619.3321881", language = "english", url = "https://www.fit.vut.cz/research/publication/11902" }