Power set construction and its variants to eliminate "dead", "unreachable" states have been the prevalent methods of minimization from Non-Deterministic Finite Automata to Deterministic Finite Automata. The drawbacks for these were the lack of adaptive improvements into the minimization process. The adaptation by means of cognitive models - perception, reasoning and decision-making - is being proposed as a better solution while handling real life problems involving decision making on transitions of states. The non-determinism added by the ambiguity in parallel transition options are resolved by applying cognitive processes. The ACT-R architecture is found to be convenient for modeling systems whose functionality can be deconstructed into a set of states, with the transition functions being modeled as production rules. The resultant model reduces the number of states from 2k to manageable values.
展开▼
