Cellular Automata Classification

Some time ago I created a small script to convert numbers into automata rules which generate different patterns. Using this I can try generating an infinite number of different patterns, the problem is that most of them are not really intereseting and I have not time to check them one by one. That's why here I try to automatize the process of finding out the intereseting cellular automatas.

Clusterization

Cellular automata patterns are clusterized based on features related to beauty such as fractal dimensionality and compression efficiency. You can read more about why to use these features here: Forsythe, Alex, et al. "Predicting beauty: fractal dimension and visual complexity in art." British journal of psychology 102.1 (2011): 49-70.

Cluster Examples

Cluster 0: Automata patterns without complexity

Cluster 1

Cluster 2

Cluster 3

Cluster 4

Cluster 5: Most complex automata patterns