论文信息 - TRAINING MANY SMALL HIDDEN MARKOV MODELS

TRAINING MANY SMALL HIDDEN MARKOV MODELS

This paper describes research in progress on two quite different ways of training systems that are composed of many small Hidden Markov Models (HMM’s). The first is a purely discriminative method in which all of the parameters of all the HMM’s are adjusted to optimize classification performance. The second is an unsupervised method in which many little HMM’s are used to model the probability density of a single set of sequences.

Geoffrey E. Hinton | Gatsby

[1] Claude E. Shannon,et al. Prediction and Entropy of Printed English , 1951 .

[2] John S. Bridle,et al. Training Stochastic Model Recognition Algorithms as Networks can Lead to Maximum Mutual Information Estimation of Parameters , 1989, NIPS.

[3] Geoffrey E. Hinton,et al. Mean field networks that learn to discriminate temporally distorted strings , 1991 .

[4] Geoffrey E. Hinton,et al. Simplifying Neural Networks by Soft Weight-Sharing , 1992, Neural Computation.

[5] Yoshua Bengio,et al. Diffusion of Credit in Markovian Models , 1994, NIPS.

[6] Terrence J. Sejnowski,et al. An Information-Maximization Approach to Blind Separation and Blind Deconvolution , 1995, Neural Computation.

[7] R. Tibshirani,et al. Discriminant Analysis by Gaussian Mixtures , 1996 .

[8] Hervé Bourlard,et al. Subband-based speech recognition , 1997, 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[9] Ronald A. Cole,et al. The CSLU speaker recognition corpus , 1998, ICSLP.

[10] Jean-François Bonastre,et al. Subband architecture for automatic speaker recognition , 2000, Signal Process..

[11] Lawrence K. Saul,et al. A statistical model for robust integration of narrowband cues in speech , 2001, Comput. Speech Lang..

[12] Geoffrey E. Hinton,et al. Products of Hidden Markov Models , 2001, AISTATS.

[13] Geoffrey E. Hinton. Training Products of Experts by Minimizing Contrastive Divergence , 2002, Neural Computation.