Abstract
Hidden Markov Network (HMnet)[1] is a kind of statisticalfinite-state automatons. Discrete-type HMnet and NL-HMnet[2] can be used as a language model in a speech recognition system. NL-HMnet has no self-loop transition. The proposed algorithms for constructing Discrete-type HMnet and NL-HMnet require the pre-defined total number of states in HMnet. The optimum number of states can be changed by various factors such as kind of task, number of training samples, and so on. It is desirable to automatically determine the optimum number of states for each condition.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
J. Takami and S. Sagayama: A Successive State Splitting Algorithm for Efficient Allophone Modeling. Proc. ICASSP’92, Vol. I, pp. 573–576, 1992.
M. Suzuki and S. Makino: An Automatic Acquisition Method of Statistic Finite-State Automaton for Sentence. Proc. ICASSP’99, Vol. II, pp. 737–740, 1999.
F. Jelinek and R. L. Mercer, “Interpolated Estimation of Markov Source Parameters from Sparse Data”, Pattern Recognition in Practice, E. S. Gelsema and L. N. Kanal, North-Holland, pp. 381–397, 1980.
J. Rissanen: Universal Coding, Information, Prediction, and Estimation. IEEE Trans. IT, Vol. 30, No. 4, pp. 629–636, 1984.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2000 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Suzuki, M., Makino, S. (2000). Automatic Determination Algorithm for the Optimum Number of States in NL-HMnet. In: Arikawa, S., Morishita, S. (eds) Discovery Science. DS 2000. Lecture Notes in Computer Science(), vol 1967. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-44418-1_36
Download citation
DOI: https://doi.org/10.1007/3-540-44418-1_36
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-41352-3
Online ISBN: 978-3-540-44418-3
eBook Packages: Springer Book Archive