Query learning of subsequential transducers

An efficient (polynomial time) algorithm is presented for the problem of learning subsequential transducers given the ability to make two kind of queries; translation queries, where the translation of a given string is returned, and equivalence queries, that are answered either positively or with a counterexample. A probabilistic setting in which equivalence queries are substituted by a random sample oracle is also studied and the corresponding modifications to the algorithm presented.

[1]  Daniel Gildea,et al.  Automatic Induction of Finite State Transducers for Simple Phonological Rules , 1995, ACL.

[2]  Enrique Vidal,et al.  Learning Subsequential Transducers for Pattern Recognition Interpretation Tasks , 1993, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Enrique Vidal,et al.  Some results with a trainable speech translation and understanding system , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[4]  Jean Berstel,et al.  Transductions and context-free languages , 1979, Teubner Studienbücher : Informatik.

[5]  E. Mark Gold,et al.  Language Identification in the Limit , 1967, Inf. Control..

[6]  Leslie G. Valiant,et al.  A theory of the learnable , 1984, STOC '84.

[7]  Dana Angluin,et al.  Learning Regular Sets from Queries and Counterexamples , 1987, Inf. Comput..