ISCA Archive ICSLP 1994
ISCA Archive ICSLP 1994

Minimum-error-rate training of predictive neural network models

KyungMin Na, JaeYeol Rheem, SouGuil Ann

Recently, predictive neural network models (PNNM) have proven successful in various speech recognition tasks. But, they suffer from poor discrimination for acoustically similar speech signals. In this paper, a new discriminative training algorithm based on the minimum-error-rate decision rule is proposed. Experiments on the Korean digits recognition have shown 37.5 % reduction of the number of recognition errors.


Cite as: Na, K., Rheem, J., Ann, S. (1994) Minimum-error-rate training of predictive neural network models. Proc. 3rd International Conference on Spoken Language Processing (ICSLP 1994), 1479-1482

@inproceedings{na94_icslp,
  author={KyungMin Na and JaeYeol Rheem and SouGuil Ann},
  title={{Minimum-error-rate training of predictive neural network models}},
  year=1994,
  booktitle={Proc. 3rd International Conference on Spoken Language Processing (ICSLP 1994)},
  pages={1479--1482}
}