DocumentCode :
290263
Title :
A new model-discriminant training algorithm for hybrid NN-HMM systems
Author :
Reichl, W. ; Caspary, P. ; Ruske, G.
Author_Institution :
Lehrstuhl fur Datenverarbeitung, Munchen Univ., Germany
Volume :
ii
fYear :
1994
fDate :
19-22 Apr 1994
Abstract :
This paper describes a hybrid system for continuous speech recognition consisting of a neural network (NN) and a hidden Markov model (HMM). The system is based on a multilayer perceptron, which approximates the a-posteriori probability of a sequence of states, derived from semi-continuous hidden Markov models. The classification is based on a total score for each hybrid model, attained from a Viterbi search on the state probabilities. Due to the unintended discrimination between the states in each model, a new training algorithm for the hybrid neural networks is presented. The utilized error function approximates the misclassification rate of the hybrid system. The discriminance between the correct and the incorrect models is optimized during the training by the `Generalized Probabilistic Descent Algorithm´, resulting in a minimum classification error. No explicit target values for the neural net output nodes are used, as in the usual backpropagation algorithm with a quadratic error function. In basic experiments up to 56% recognition rate were achieved on a vowel classification task and up to 69% on a consonant cluster classification task
Keywords :
backpropagation; feedforward neural nets; hidden Markov models; multilayer perceptrons; probability; search problems; speech recognition; Viterbi search; backpropagation algorithm; consonant cluster classification; continuous speech recognition; error function; generalized probabilistic descent algorithm; hidden Markov model; hybrid NN-HMM systems; misclassification rate; model-discriminant training algorithm; multilayer perceptron; neural net output nodes; neural network; quadratic error function; recognition rate; semi-continuous hidden Markov models; state probabilities; training algorithm; vowel classification; Backpropagation algorithms; Clustering algorithms; Error correction; Hidden Markov models; Iterative algorithms; Multilayer perceptrons; Neural networks; Production systems; Speech recognition; Viterbi algorithm;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1994. ICASSP-94., 1994 IEEE International Conference on
Conference_Location :
Adelaide, SA
ISSN :
1520-6149
Print_ISBN :
0-7803-1775-0
Type :
conf
DOI :
10.1109/ICASSP.1994.389565
Filename :
389565
Link To Document :
بازگشت