Non-stationary Mixture Research Articles

Double Nonstationarity: Blind Extraction of Independent Nonstationary Vector/Component from Nonstationary Mixtures—Performance Analysis

Double Nonstationarity: Blind Extraction of Independent Nonstationary Vector/Component from Nonstationary Mixtures—Performance Analysis

Pseudo-Bayesian Approach for Robust Mode Detection and Extraction Based on the STFT

This paper addresses the problem of disentangling nonoverlapping multicomponent signals from their observation being possibly contaminated by external additive noise. We aim to extract and to retrieve the elementary components (also called modes) present in an observed nonstationary mixture signal. To this end, we propose a new pseudo-Bayesian algorithm to perform the estimation of the instantaneous frequency of the signal modes from their time-frequency representation. In a second time, a detection algorithm is developed to restrict the time region where each signal component behaves, to enhance quality of the reconstructed signal. We finally deal with the presence of noise in the vicinity of the estimated instantaneous frequency by introducing a new reconstruction approach relying on nonbinary band-pass synthesis filters. We validate our methods by comparing their reconstruction performance to state-of-the-art approaches through several experiments involving both synthetic and real-world data under different experimental conditions.

Double Nonstationarity: Blind Extraction of Independent Nonstationary Vector/Component From Nonstationary Mixtures—Algorithms

In this article, nonstationary mixing and source models are combined for developing new fast and accurate algorithms for Independent Component or Vector Extraction (ICE/IVE), one of which stands for a new extension of the well-known FastICA. This model allows for a moving source-of-interest (SOI) whose distribution on short intervals can be (non-)circular (non-)Gaussian. A particular Gaussian source model assuming tridiagonal covariance matrix structures is proposed. It is shown to be beneficial in the frequency-domain speaker extraction problem. The algorithms are verified in simulations. In comparison to the state-of-the-art algorithms, they show superior performance in terms of convergence speed and extraction accuracy.

Quantum fluids of light in all-optical scatterer lattices

One of the recently established paradigms in condensed matter physics is examining a system’s behaviour in artificial potentials, giving insight into phenomena of quantum fluids in hard-to-reach settings. A prominent example is the matter-wave scatterer lattice, where high energy matter waves undergo transmission and reflection through narrow width barriers leading to stringent phase matching conditions with lattice band formation. In contrast to evanescently coupled lattice sites, the realisation of a scatterer lattice for macroscopic matter-wave fluids has remained elusive. Here, we implement a system of exciton-polariton condensates in a non-Hermitian Lieb lattice of scatterer potentials. By fine tuning the lattice parameters, we reveal a nonequilibrium phase transition between distinct regimes of polariton condensation: a scatterer lattice of gain guided polaritons condensing on the lattice potential maxima, and trapped polaritons condensing in the potential minima. Our results pave the way towards unexplored physics of non-Hermitian fluids in non-stationary mixtures of confined and freely expanding waves.

Storm characterization and simulation for damage evolution models of maritime structures

This paper presents a new approach to statistically characterize and simulate the wave climate under storm conditions. The methodology includes the joint selection of the parameters that identify storm events (significant wave height threshold, minimum storm duration and minimum interarrival time between consecutive storms) by means of hypothesis testing on the distribution functions of the number of storm events and the elapsing time between storms, providing an improved characterization of the parameters that define storm events. The main wave variables and their temporal dependence are characterized by non-stationary mixture distribution functions and a vector autoregressive model. This allows to adequately reproduce the random temporal evolution of storm events, crucial for the study of damage progression in maritime structures without the use of predefined geometries. The long-term time series of storm events and calm periods is obtained using copula functions which analyze the joint dependence of storm duration and interarrival time for separate climate intervals. The model is applied to hindcast data at a location of the Mediterranean sea close to the Granada coast in Spain to show its ability to reproduce wave storm conditions accounting for the time variability of the storminess. An example of application, using a large number of simulations and a damage progression model in a maritime structure, is presented.

DIFFEOMORPHIC POINT SET REGISTRATION USING NON-STATIONARY MIXTURE MODELS.

This paper investigates a diffeomorphic point-set registration based on non-stationary mixture models. The goal is to improve the non-linear registration of anatomical structures by representing each point as a general non-stationary kernel that provides information about the shape of that point. Our framework generalizes work done by others that use stationary models. We achieve this by integrating the shape at each point when calculating the point-set similarity and transforming it according to the calculated deformation. We also restrict the non-rigid transform to the space of symmetric diffeomorphisms. Our algorithm is validated in synthetic and human datasets in two different applications: fiber bundle and lung airways registration. Our results shows that non-stationary mixture models are superior to Gaussian mixture models and methods that do not take into account the shape of each point.

Does Experience Teach? Professionals and Minimax Play in the Lab

Does expertise in strategic behavior obtained in the field transfer to the abstract setting of the laboratory? Palacios-Huerta and Volij (2008) argued that the behavior of professional soccer players in mixed-strategy games conforms closely to minimax play, while the behavior of students (who are presumably novices in strategic situations requiring unpredictability) does not. We reexamine their data, showing that the play of professionals is inconsistent with the minimax hypothesis in several respects: (i) professionals follow nonstationary mixtures, with action frequencies that are negatively correlated between the first and the second half of the experiment, (ii) professionals tend to switch between under- and overplaying an action relative to its equilibrium frequency, and (iii) the distribution of action frequencies across professionals is far from the distribution implied by minimax. In each respect, the behavior of students conforms more closely to the minimax hypothesis.

Asymptotic normality of the Nadaraya–Watson estimator for nonstationary functional data and applications to telecommunications

We study a nonparametric regression model, where the explanatory variable is nonstationary dependent functional data and the response variable is scalar. Assuming that the explanatory variable is a nonstationary mixture of stationary processes and general conditions of dependence of the observations (implied in particular by weak dependence), we obtain the asymptotic normality of the Nadaraya–Watson estimator. Under some additional regularity assumptions on the regression function, we obtain asymptotic confidence intervals for the regression function. We apply this result to estimate the quality of service for an end-to-end connection on a network.

WITHDRAWN: Separation of non-stationary mixtures of cross-correlated processes using sequential Monte Carlo

WITHDRAWN: Separation of non-stationary mixtures of cross-correlated processes using sequential Monte Carlo

Separation of combustion noise and piston-slap in diesel engine—Part I: Separation of combustion noise and piston-slap in diesel engine by cyclic Wiener filtering

The main purpose of this study is to characterize the relative noise given out by a diesel engine, around the Top Dead Centre (TDC) by quantifying the proportions of “mechanical noise” originating mainly from piston-slap on the one hand and ‘thermal noise” originating from combustion on the other hand. Two different approaches are described here to solve this problem. In the first part of the paper, the cylinder pressure is measured and used as a reference in order to reconstruct the thermal noise. Next, we propose a method based on applying a cyclic Wiener filter to the measured cylinder pressure in order to separate the noises of mechanical and thermal origins. The final result is to reduce the engine resulting noise. The second part of the paper is devoted to blind source separation (BSS) methods applied on signals issued from accelerometers placed on one of the cylinders. It develops a BSS method based on a convolutive model of non-stationary mixtures and introduces a new method based on the joint diagonalization of time varying spectral matrices of the observations. Both methods are then applied to real data and the estimated sources are finally validated by several physical arguments.

Separation of combustion noise and piston-slap in diesel engine—Part II: Separation of combustion noise and piston-slap using blind source separation methods

This paper is devoted to blind separation of combustion noise and piston-slap in diesel engines. The two phenomena are recovered only from signals issued from accelerometers placed on one of the cylinders. A blind source separation (BSS) method is developed, based on a convolutive model of non-stationary mixtures. We introduce a new method based on the joint diagonalisation of the time varying spectral matrices of the observation records and a new technique to handle the problem of permutation ambiguity in the frequency domain. This method is then applied to real data and the estimated sources are validated by several physical arguments. So, the contribution of the piston-slap and the combustion noise can be recovered for all the sensors. The energy of the two phenomena can then be given with regards to the position of the accelerometers.