Bayesian inference for finite-state transducers

We describe a Bayesian inference algorithm that can be used to train any cascade of weighted finite-state transducers on end-to-end data. We also investigate the problem of automatically selecting from among multiple training runs. Our experiments on four different tasks demonstrate the genericity of this framework, and, where applicable, large improvements in performance over EM. We also show, for unsupervised part-of-speech tagging, that automatic run selection gives a large improvement over previous Bayesian approaches.