Training Mixture Density HMMs with SOM and LVQ

Training Mixture Density HMMs with SOM and LVQ (1997)

Cached

Download Links

[www.cis.hut.fi]

by Mikko Kurimo , Mikko Kurimo , Mikko Kurimo

Citations:

4 - 2 self

BibTeX

@MISC{Kurimo97trainingmixture,
    author = {Mikko Kurimo and Mikko Kurimo and Mikko Kurimo},
    title = {Training Mixture Density HMMs with SOM and LVQ},
    year = {1997}
}

Bookmark

OpenURL

Abstract

¯ The objective of this paper is to present experiments and discussions of how some neural network algorithms can help the phoneme recognition with mixture density hidden Markov models (MDHMMs). In MDHMMs the modeling of the stochastic observation processes associated with the states is based on the estimation of the probability density function of the short-time observations in each state as a mixture of Gaussian densities. The Learning Vector Quantization (LVQ) is used to increase the discrimination between dioeerent phoneme models both during the initialization of the Gaussian codebooks and during the actual MDHMM training. The Self-Organizing Map (SOM) is applied to provide a suitably smoothed mapping of the training vectors to accelerate the convergence of the actual training. The obtained codebook topology can also be exploited in the recognition phase to speed up the calculations to approximate the observation probabilities. The experiments with LVQ and SOMs show reductions both...

Citations

3116	A tutorial on hidden Markov models and selected applications in speech recognition - Rabiner - 1989
2726	The self-organizing map - Kohonen - 1990
599	The Viterbi algorithm - Forney - 1973
357	Multivariable functional interpolation and adaptive networks,Complex - Lowe
270	Statistical inference for probabilistic functions of finite state Markov chains - Baum, Petrie - 1966
156	Some statistical issues in the comparison of speech recognition algorithms - Gillick, Cox - 1989
152	Continuous speech recognition by statistical methods - Jelinek - 1976
142	Discriminative Learning for Minimum Error Classification - Juang, Katagiri - 1992
123	R.: Maximum mutual information estimation of hidden Markov model parameters for speech recognition - Bahl, Brown, et al. - 2000
88	Links between Markov Models and Multilayer Perceptrons - Bourlard, Wellekens - 1989
88	The “neural” phonetic typewriter - Kohonen - 1988
79	Maximum likelihood estimation for multivariate observations of Markov sources - Liporace - 1982
77	Review of neural networks for speech recognition - Lippmann
75	The segmental Kmeans algorithm for estimating parameters of hidden Markov models - Juang, Rabiner - 1990
63	The DRAGON system – an overview - Baker - 1975
58	Maximum likelihood estimation for mixture multivariate observations of Markov chains - Juang - 1985
58	Hidden Markov models for speech recognition - Juang, Rabiner - 1990
55	LVQ PAK: A program package for the correct application of Learning Vector Quantization algorithms - Kohonen, Kangas, et al.
54	Tied mixture continuous parameter modeling for speech recognition - Bellegarda - 1990
51	Semi-continuous hidden Markov models for speech signals - Huang, Jack - 1989
39	A segmental K-means training procedure for connected word recognition - Rabiner, Wilpon, et al. - 1986
38	Learning vector quantization for pattern recognition - Kohonen - 1989
35	New discriminative training algorithms based on the generalised descent method - Katagiri, Lee, et al. - 1991
34	GENONES: Optimizing the Degree of Mixture Tying in a Large Vocabulary - Digalakis, Murveit - 1994
31	Segmental GPD Training Of HMM Based Speech Recognizer - Chou, Juang, et al. - 1992
23	Improving Speech Recognition Performance via - Hwang, Rosenfeld, et al. - 1994
22	On a model-robust training method for speech recognition - Nádas, Nahamoo, et al. - 1988
22	Combining Hidden Markov Models and Neural Network Classifiers - Niles, Silverman - 1990
18	A new algorithm for the estimation of hidden Markov model parameters - Bahl, Brown, et al. - 1988
14	A Hybrid Speech Recognition System using HMMS with an LVQ-Trained Codebook, ICASSP - Iwamida, Katagiri, et al. - 1990
11	The Lincoln Robust Continuous Speech Recogniser - Paul - 1989
11	Status report of the Finnish phonetic typewriter project - Torkkola, Kangas, et al. - 1991
10	A Speech Recognizer Using Radial Basis Function Neural Nerworks in an HMM Framework - Singer, Lippmann - 1992
9	An optimal discriminative training method for continuous mixture density HMMs - Mizuta, Nakajima - 1990
8	A theory of adaptive pattern classi ers - Amari - 1967
7	Training continuous density hidden Markov models in association with self-organizing maps and LVQ - Kurimo, Torkkola - 1992
6	et.al., A New hybrid algorithm for speech recognition based on HMM segmentation and learning Vector quantization - Katagiri
6	Hybrid training method for tied mixture density hidden Markov models using Learning Vector Quantization and Viterbi estimation - Kurimo - 1994
6	Using the self-organizing map to speed up the probability density estimation for speech recognition with mixture density HMMs - Kurimo, Somervuo - 1996
5	Adaptation algorithms for large scale HMM recognizers - Zavaliagkos, Schwartz, et al. - 1995
4	GPD training of dynamic programmingbased speech recognizers - Komori, Katagiri - 1992
4	Corrective tuning by applying LVQ for continuous density and semi-continuous Markov models - Kurimo - 1994
4	Segmental LVQ3 training for phoneme-wise tied mixture density HMMs - Kurimo - 1996
4	Combining LVQ with continuous density hidden Markov models in speech recognition - Kurimo, Torkkola - 1992
4	Using multiple vector quantization and semicontinuous hidden Markov models for speech recognition - Peinado, Segura, et al. - 1994
3	Using LVQ to enhance semi-continuous hidden Markov models for phonemes - Kurimo - 1993
3	Smoothing HMMs by means of a SOM - Monte - 1992
2	An HMM/MLP architecture for sequence recognition - Cho, Kim - 1995
2	A competitive algorithm for training HMM for speech recognition - Galindo - 1995
2	Multi-dimensional HMM parameter estimation using self-organizing feature map for speech recognition - Kim, Lee, et al. - 1994