A method of regular grammar inference in computational learning for classification problems is presented. We classify the strings by generating multiple subclasses. A construction algorithm is presented for the classification automata from positive classified examples. We prove the correctness of the algorithm, and suggest some possible extensions.
展开▼