Paper: Parameter Estimation For Probabilistic Finite-State Transducers

ACL ID P02-1001
Title Parameter Estimation For Probabilistic Finite-State Transducers
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2002
Authors

Weighted finite-state transducers suffer from the lack of a train- ing algorithm. Training is even harder for transducers that have been assembled via finite-state operations such as composition, minimization, union, concatenation, and closure, as this yields tricky parameter tying. We formulate a “parameterized FST” paradigm and give training algorithms for it, including a gen- eral bookkeeping trick (“expectation semirings”) that cleanly and efficiently computes expectations and gradients. 1 Background and Motivation Rational relations on strings have become wide- spread in language and speech engineering (Roche and Schabes, 1997). Despite bounded memory they are well-suited to describe many linguistic and tex- tual processes, either exactly or approximately. A relation is a set ...