EUROSPEECH 2003 - INTERSPEECH 2003
We present a method for conditional maximum likelihood estimation of N-gram models used for text or speech utterance classification. The method employs a well known technique relying on a generalization of the Baum-Eagon inequality from polynomials to rational functions. The best performance is achieved for the 1-gram classifier where conditional maximum likelihood training reduces the class error rate over a maximum likelihood classifier by 45% relative.
Bibliographic reference. Chelba, Ciprian / Acero, Alex (2003): "Discriminative training of n-gram classifiers for speech and text routing", In EUROSPEECH-2003, 2777-2780.