In this paper, we describe new techniques to signi cantly reduce computational, storage and memory access requirements of a tied-mixture HMM based speech recognition system. Although continuous mixture HMMs o er improved recognition performance, we show that tied-mixture HMMs may o er signi cant advantage in complexity reduction for low-cost implementations. In particular, we consider two tasks...

The one–dimensional differential equations for the conservation mixture mass and mixture momentum coupled by a relative velocity conservation equation have been solved for an isentropic mixture of gas and liquid two–phase flow. Under certain restrictions the resulting system of partial differential equations is hyperbolic and allow discontinuous solutions. These equations have been solved by th...

We deduce an overcomplete free energy functional for D=1 particle systems with next neighbor interactions, where the set of redundant variables are the local block densities +i of i interacting particles. The idea is to analyze the decomposition of a given pure system into blocks of i interacting particles by means of a mapping onto a hard rod mixture. This mapping uses the local activity of co...

Density estimation, especiallymultivariate density estimation, is a fundamental problem in nonparametric inference. In the Bayesian approach, Dirichlet mixture priors are often used in practice for such problems. However, the asymptotic properties of such priors have only been studied in the univariate case. We extend the L1-consistency of Dirichlet mixutures in the multivariate density estimat...

The purpose of this research is to investigate a zeotropic working fluid mixture in terms of its performance in an organic Rankine cycle (ORC) and the heat transfer characteristics in an air-cooled condenser (ACC). The motivation of this study is that it is well known that the use of a mixture improves the efficiency of an ORC system. However, the behaviour of the mixture in the condensers is n...

Recent developments in system identification have brought attention to regularized kernel-based methods. This type of approach has been proven to compare favorably with classic parametric methods. However, current formulations are not robust with respect to outliers. In this paper, we introduce a novel method to robustify kernel-based system identification methods. To this end, we model the out...

A Bayesian filtering algorithm is developed for a class of state-space systems that can be modelled via Gaussian mixtures. In general, the exact solution to this filtering problem involves an exponential growth in the number of mixture terms and this is handled here by utilising a Gaussian mixture reduction step after both the time and measurement updates. In addition, a square-root implementat...

This paper describes the development of an approximate method for propagating uncertainty through stochastic dynamical systems using a quadrature rule integration based method. The development of quadrature rules for Gaussian mixture distributions is discussed. A numerical solution to this problem is considered that uses a Gram-Schmidt process. Simulation results are presented where the quadrat...