A Model Reduction Algorithm for Hidden Markov Models

This paper presents a two step model reduction algorithm for discrete-time, finite state, finite alphabet hidden Markov models. The complexity measure used is the cardinality of the state space of the underlying Markov chain. In the first step, hidden Markov models are associated with a certain class of stochastic jump linear systems, namely the ones where the parametric input is a sequence of independent identically distributed random variables. The image of the high dimensional hidden Markov model in this class of stochastic jump linear systems is simplified by means of a balanced truncation algorithm, which was developed by Kotsalis (2006). Subsequently, the reduced order stochastic jump linear system is modified, so that it meets the constraints of an image of a hidden Markov model of the same order. This is achieved by solving a low dimensional non convex optimization problem. Numerical simulation results provide evidence that the proposed algorithm computes accurate reduced order hidden Markov models, while achieving a compression of the state space by orders of magnitude

[1]  Lawrence R. Rabiner,et al.  A tutorial on hidden Markov models and selected applications in speech recognition , 1989, Proc. IEEE.

[2]  Cao Xiao-yan.  Markov Models in Bioinformatics , 2000 .

[3]  Peter Spreij,et al.  4Approximate realization of hidden Markov chains , 2002, Proceedings of the IEEE Information Theory Workshop.

[4]  Eduardo D. Sontag,et al.  Realization Theory of Discrete-Time Nonlinear Systems: Part I - The Bounded Case , 1979 .

[5]  A. Heller.  On Stochastic Processes Derived From Markov Chains , 1965 .

[6]  Mariëlle Stoelinga,et al.  An Introduction to Probabilistic Automata , 2002, Bull. EATCS.

[7]  M. Vidyasagar,et al.  The Realization Problem for Hidden Markov Models: The Complete Realization Problem , 2005, Proceedings of the 44th IEEE Conference on Decision and Control.

[8]  D. Blackwell,et al.  On the Identifiability Problem for Functions of Finite Markov Chains , 1957 .

[9]  Samuel Eilenberg,et al.  Automata, languages, and machines. A , 1974, Pure and applied mathematics.

[10]  V. Blondel,et al.  The presence of a zero in an integer linear recurrent sequence is NP-hard to decide , 2002 .

[11]  T. Koski.  Hidden Markov Models for Bioinformatics , 2001 .

[12]  Luca Benvenuti,et al.  A tutorial on the positive realization problem , 2004, IEEE Transactions on Automatic Control.

[13]  R. P. Marques,et al.  Discrete-Time Markov Jump Linear Systems , 2004, IEEE Transactions on Automatic Control.

[14]  Brian D. O. Anderson,et al.  The Realization Problem for Hidden Markov Models , 1999, Math. Control. Signals Syst..

[15]  A. Isidori.  Direct construction of minimal bilinear realizations from nonlinear input-output maps , 1973 .

[16]  Carolyn L. Beck.  On formal power series representations for uncertain systems , 2001, IEEE Trans. Autom. Control..

[17]  Giorgio Picci,et al.  On the Internal Structure of Finite-State Stochastic Processes , 1978 .

[18]  Berndt Farwer,et al.  ω-automata , 2002 .

[19]  Robert E. Mahony,et al.  Lumpable hidden Markov models-model reduction and reduced complexity filtering , 2000, IEEE Trans. Autom. Control..

[20]  Jean Berstel,et al.  Rational series and their languages , 1988, EATCS monographs on theoretical computer science.

[21]  Sang Joon Kim,et al.  A Mathematical Theory of Communication , 2006 .

[22]  R. Bhar,et al.  Hidden Markov Models: Applications to Financial Economics , 2004 .

[23]  Thomas Sudkamp,et al.  Languages and Machines , 1988 .

[24]  Georgios Kotsalis.  Model reduction for Hidden Markov models , 2006 .

[25]  Paavo Turakainen,et al.  Generalized automata and stochastic languages , 1969 .