This thesis presents the results and findings obtained in developing multiple modules for the CAM-brain machine (CBM), partially evaluating its proficiency as a pattern recognizer. The CBM is a field programmable gate array (FPGA)-based device, which implements a genetic algorithm (GA) to evolve a 3D cellular automata (CA)-based neural network module. The first set of modules evolved in this work for the CBM was for the recognition of a single character of the alphabet, with all 26 characters being mapped onto the CA cubic space. The next set of modules evolved was for the recognition of a word, mapping a sequence of alphabetic characters onto the CA cubic space. For each set of modules noise was added both statically and dynamically, mimicking the behavior of an analog signal. Each module was run using the simulation software, determining the proficiency of the CBM by a measure of fitness. |