Comparative study on Null Boundary and Periodic Boundary 3-neighborhood Multiple Attractor Cellular Automata for classification

Abstract

This paper reports a generic analysis on null boundary and periodic boundary 3-neighborhood multiple attractor cellular automata (MACA) for showing the comparative study in classification technique. Cellular automata (CA) is now-a-days an essential tool for researchers in the area of pattern recognition, pattern generation, testing field, fault diagnosis and so on. So, general knowledge on CA up to some extent is a must for researchers in these areas. A CA may be linear or non-linear in behavior. A linear/additive CA employs XOR/XNOR logic, while a non-linear CA employs AND/OR/NOT logic. This paper shows a graph analysis along with state transition behavior of CA cells. A rule vector graph (RVG) is generated from the rule vector (RV) of a CA. Linear time algorithms are reported for generation of RVG. MACA provides an implicit memory to store the patterns. Search operation to identify the class of a pattern out of several classes boils down to running a CA for one time step. This demands storage of RV and seed values. MACA is based on sound theoretical foundation of CA technology. This paper only concentrates on MACA since it is responsible for classifying the various types of patterns.