A General Strategy for Hidden Markov Chain Parameterisation in Composite Feature-Spaces
A general technique for the construction of hidden Markov models (HMMs) from multiple-variable time-series observations in noisy experimental environments is set out. The proposed methodology provides an ICA-based feature-selection technique for determining the number, and the transition sequence, of underlying hidden states, along with the statistics of the observed-state emission characteristics. In retaining correlation information between features, the method is potentially far more general than Gaussian mixture model HMM parameterisation methods such as Baum-Welch re-estimation, to which we demonstrate our method reduces when an arbitrary separation of features, or an experimentally-limited feature-space is imposed.
KeywordsHide Markov Model Independent Component Analysis Gaussian Mixture Model Independent Component Analysis Hide State
- 1.Baum, L., Eagon, J.: American Mathematical Society Bulletin, vol. 73, pp. 360–363 (1967)Google Scholar
- 2.Young, S., Odell, J., Ollason, D., Valtchev, V., Woodland, P.: The HTK Book for HTK Version 2.1. Cambridge University, Cambridge (1997)Google Scholar
- 4.Comon, P.: Independent component analysis, A new concept? IEEE Signal Processing Mag. 36(3) (1994)Google Scholar
- 6.Diaconis, P., Freedman, D.: Asymptotics of graphical projection pursuit. Ann. Stat. 12, 793–815 Google Scholar
- 7.Windridge, D., Bowden, R., Kittler, J. (Univ. of Surrey Technical Report: VSSP-TR- 3/2004), UNIS, UKGoogle Scholar