We consider the use of language models whose size and accuracy are intermediate between different order n-gram models. Two types of models are studied in particular. Aggregate Markov models are classbased bigram models in which the mapping from words to classes is probabilistic. Mixed-order Markov models combine bigram models whose predictions are conditioned on different words. Both types of m...

First-order Markov models have been successfully applied to many problems, for example in modeling sequential data using Markov chains, and modeling control problems using the Markov decision processes (MDP) formalism. If a first-order Markov model’s parameters are estimated from data, the standard maximum likelihood estimator considers only the first-order (single-step) transitions. But for ma...

Abstract: A reversible jump Markov chain Monte Carlo technique is applied to estimate the number and parameters of peaks in ubiquitous physical problems in the framework of Bayesian probability theory. For measured physical spectra often only the functional form of the structures is known but the number of the peaks and the parameters are unknown. The full joint posterior distribution for all p...

this paper proposes a new method for parameter estimation of distorted single phase signals, through an improved demodulation-based phase tracking incorporated with a frequency adaptation mechanism. the simulation results demonstrate the superiority of the proposed method compared to the conventional sogi (second-order generalized integrator)-based approach, in spite of the dc-offset and harmon...

A hidden Markov model method for estimating an a posteriori distribution of the amplitude of communications signals is presented. As the signal to noise ratio decreases the hidden Markov model method is shown to perform significantly better than a conventional histogram method for characterising the amplitude distribution. The HMM estimation is performed within a Expectation Maximisation method...

Markov switching GARCH models have been developed in order to address the statistical regularity observed in financial time series such as strong persistence of conditional variance. However, Maximum Likelihood Estimation faces a implementation problem since the conditional variance depends on all the past history of state. This paper shows that this problem can be handled easily in Bayesian in...