Improving the IBM Alignment Models Using Variational Bayes

Abstract

Bayesian approaches have been shown to reduce the amount of overfitting that occurs when running the EM algorithm, by placing prior probabilities on the model parameters. We apply one such Bayesian technique, variational Bayes, to the IBM models of word alignment for statistical machine translation. We show that using variational Bayes improves the… (More)

6 Figures and Tables

Topics

  • Presentations referencing similar topics