5th European Conference on Speech Communication and Technology

Rhodes, Greece
September 22-25, 1997

Comparison of Optimization Methods for Discriminative Training Criteria

Ralf Schlüter, W. Macherey, S. Kanthak, Hermann Ney, Lutz Welling

Lehrstuhl fur Informatik VI RWTH Aachen, Germany

In this work we compare two parameter optimization techniques for discriminative training using the MMI criterion: the extended Baum- Welch (EBW) algorithm and the generalized probabilistic descent (GPD) method. Using Gaussian emission densities we found special expressions for the step sizes in GPD, leading to reestimation formula very similar to those derived for the EBW algorithm. Results were produced for both the TI digitstring and the SieTill corpus for continuously spoken American English and German digitstrings. The results for both techniques do not show significant differences. This experimental results support the strong link between EBW and GPD as expected from the analytic comparison.

Full Paper

Bibliographic reference.  Schlüter, Ralf / Macherey, W. / Kanthak, S. / Ney, Hermann / Welling, Lutz (1997): "Comparison of optimization methods for discriminative training criteria", In EUROSPEECH-1997, 15-18.