Training and extraction of fuzzy finite state automata in recurrent neural networks

We present a machine learning approach for the extraction of fuzzy finite-state automata (FFAs) from recurrent neural networks. After successful training on strings with fuzzy membership t c [0,I], we apply a generalisation of the Trakhtenbrot-Barzdin algorithm to extract FFAs in symbolic form from the trained network using the string labelling assigned by the trained network. Our results demonstrate that the approach successfully extracts the correct deterministic equivalent automata for strings much shorter than the longest string in the training set.