Improved A Posteriori Speech Presence Probability Estimation Based on a Likelihood Ratio With Fixed Priors

In this paper, we present an improved estimator for the speech presence probability at each time-frequency point in the short-time Fourier transform domain. In contrast to existing approaches, this estimator does not rely on an adaptively estimated and thus signal-dependent a priori signal-to-noise ratio estimate. It therefore decouples the estimation of the speech presence probability from the estimation of the clean speech spectral coefficients in a speech enhancement task. Using both a fixed a priori signal-to-noise ratio and a fixed prior probability of speech presence, the proposed a posteriori speech presence probability estimator achieves probabilities close to zero for speech absence and probabilities close to one for speech presence. While state-of-the-art speech presence probability estimators use adaptive prior probabilities and signal-to-noise ratio estimates, we argue that these quantities should reflect true a priori information that shall not depend on the observed signal. We present a detection theoretic framework for determining the fixed a priori signal-to-noise ratio. The proposed estimator is conceptually simple and yields a better tradeoff between speech distortion and noise leakage than state-of-the-art estimators.     

Robust fusion time‐varying Kalman estimators for multisensor networked systems with mixed uncertainties

This paper addresses the problem of designing robust fusion time‐varying Kalman estimators for a class of multisensor networked systems with mixed uncertainties including multiplicative noises, missing measurements, packet dropouts, and uncertain‐variance linearly correlated measurement and process white noises. By the augmented approach, the original system is converted into a stochastic parameter system with uncertain noise variances. Furthermore, applying the fictitious noise approach, the original system is converted into one with constant parameters and uncertain noise variances. According to the minimax robust estimation principle, based on the worst‐case system with the conservative upper bounds of the noise variances, the five robust fusion time‐varying Kalman estimators (predictor, filter, and smoother) are presented by using a unified design approach that the robust filter and smoother are designed based on the robust Kalman predictor, which include three robust weighted state fusion estimators with matrix weights, diagonal matrix weights, and scalar weights, a modified robust covariance intersection fusion estimator, and robust centralized fusion estimator. Their robustness is proved by using a combination method, which consists of Lyapunov equation approach, augmented noise approach, and decomposition approach of nonnegative definite matrix, such that their actual estimation error variances are guaranteed to have the corresponding minimal upper bounds for all admissible uncertainties. The accuracy relations among the robust local and fused time‐varying Kalman estimators are proved. A simulation example is shown with application to the continuous stirred tank reactor system to show the effectiveness and correctness of the proposed results.     

Robust normal estimation and region growing segmentation of infrastructure 3D point cloud models

Modern remote sensing technologies such as three-dimensional (3D) laser scanners and image-based 3D scene reconstruction are in increasing demand for applications in civil infrastructure design, maintenance, operation, and as-built construction verification. The complex nature of the 3D point clouds these technologies generate, as well as the often massive scale of the 3D data, make it inefficient and time consuming to manually analyze and manipulate point clouds, and highlights the need for automated analysis techniques. This paper presents one such technique, a new region growing algorithm for the automated segmentation of both planar and non-planar surfaces in point clouds. A core component of the algorithm is a new point normal estimation method, an essential task for many point cloud processing algorithms. The newly developed estimation method utilizes robust multivariate statistical outlier analysis for reliable normal estimation in complex 3D models, considering that these models often contain regions of varying surface roughness, a mixture of high curvature and low curvature regions, and sharp features. An adaptation of Mahalanobis distance, in which the mean vector and covariance matrix are derived from a high-breakdown multivariate location and scale estimator called Deterministic MM-estimator (DetMM) is used to find and discard outlier points prior to estimating the best local tangent plane around any point in a cloud. This approach is capable of more accurately estimating point normals located in highly curved regions or near sharp features. Thereafter, the estimated point normals serve a region growing segmentation algorithm that only requires a single input parameter, an improvement over existing methods which typically require two control parameters. The reliability and robustness of the normal estimation subroutine was compared against well-known normal estimation methods including the Minimum Volume Ellipsoid (MVE) and Minimum Covariance Determinant (MCD) estimators, along with Maximum Likelihood Sample Consensus (MLESAC). The overall region growing segmentation algorithm was then experimentally validated on several challenging 3D point clouds of real-world infrastructure systems. The results indicate that the developed approach performs more accurately and robustly in comparison with conventional region growing methods, particularly in the presence of sharp features, outliers and noise.     

An Efficient Method To Estimate Bagging's Generalization Error

Bagging (Breiman, 1994a) is a technique that tries to improve a learning algorithm's performance by using bootstrap replicates of the training set (Efron & Tibshirani, 1993, Efron, 1979). The computational requirements for estimating the resultant generalization error on a test set by means of cross-validation are often prohibitive, for leave-one-out cross-validation one needs to train the underlying algorithm on the order of m times, where m is the size of the training set and is the number of replicates. This paper presents several techniques for estimating the generalization error of a bagged learning algorithm without invoking yet more training of the underlying learning algorithm (beyond that of the bagging itself), as is required by cross-validation-based estimation. These techniques all exploit the bias-variance decomposition (Geman, Bienenstock & Doursat, 1992, Wolpert, 1996). The best of our estimators also exploits stacking (Wolpert, 1992). In a set of experiments reported here, it was found to be more accurate than both the alternative cross-validation-based estimator of the bagged algorithm's error and the cross-validation-based estimator of the underlying algorithm's error. This improvement was particularly pronounced for small test sets. This suggests a novel justification for using bagging—more accurate estimation of the generalization error than is possible without bagging.     

Sliding-mode estimators for a class of non-linear systems affected by bounded disturbances

The problem of state estimation for a class of non-linear systems with Lipschitz non-linearities is addressed using sliding-mode estimators. Stability conditions have been found to guarantee asymptotic convergence to zero of the estimation error in the absence of noise and non-divergence if the state perturbations and measurement noise are bounded. A method is proposed to find a suitable solution to the algebraic Riccati equation on which the design of the estimator is based. The performance of the resulting sliding-mode filter minimizes an upper bound on the asymptotic estimation error. Based on such an approach, a sliding-mode estimator may be designed so as to outperform the extended Kalman filter in practical applications with models affected by uncertainty and strong, possibly unknown non-linearities, as shown by means of simulations.     

Robust weighted fusion Kalman estimators for multisensor systems with multiplicative noises and uncertain‐covariances linearly correlated white noises

This paper addresses the design of robust weighted fusion Kalman estimators for a class of uncertain multisensor systems with linearly correlated white noises. The uncertainties of the systems include the same multiplicative noises perturbations both on the systems state and measurement output and the uncertain noise variances. The measurement noises and process noise are linearly correlated. By introducing two fictitious noises, the system under consideration is converted into one with only uncertain noise variances. According to the minimax robust estimation principle, based on the worst‐case systems with the conservative upper bounds of the noise variances, the four robust weighted fusion time‐varying Kalman estimators are presented in a unified framework, which include three robust weighted state fusion estimators with matrix weights, diagonal matrix weights, scalar weights, and a modified robust covariance intersection fusion estimator. The robustness of the designed fusion estimators is proved by using the Lyapunov equation approach such that their actual estimation error variances are guaranteed to have the corresponding minimal upper bounds for all admissible uncertainties. The accuracy relations among the robust local and fused time‐varying Kalman estimators are proved. The corresponding robust local and fused steady‐state Kalman estimators are also presented, a simulation example with application to signal processing to show the effectiveness and correctness of the proposed results. Copyright © 2016 John Wiley & Sons, Ltd.     

On a Method of Empirical Risk Minimization

We consider the problem of estimating an unknown vector observed in a simple white Gaussian noise model. For the estimation, a family of projection estimators is used; the problem is to choose, based on observations, the best estimator within this family. The paper studies a method for choosing a projection estimator, based on the principle of penalized empirical risk minimization. For this estimation method, nonasymptotic inequalities controlling its quadratic risk are given.     

Robust state estimation for a class of uncertain time-delay systems

This paper considers a robust state estimation problem for a class of uncertain time-delay systems. In this problem, the noise and uncertainty are modelled deterministically via an integral quadratic constraint. The robust state estimation problem involves constructing the set of all possible states at the current time consistent with given output measurements and the integral quadratic constraint. This set is found to be an ellipsoid which is constructed via a linear state estimator.     

Estimation of the autoregressive parameters of a mixed autoregressive moving-average time series

The problem of estimating the autoregressive parameters of a mixed autoregressive moving-average (ARMA) time series (of known order) using the output data alone is treated. This problem is equivalent to the estimation of the denominator terms of the scalar transfer function of a stationary, linear discrete time system excited by an unobserved unenrrelated sequence input by employing only the observations of the scalar output. The solution of this problem solves the problem of the identification of the dynamics of a white-noise excited continuous-time linear stationary system using sampled data. The latter problem was suggested by Bartlett in 1946. The problem treated here has appeared before in the engineering literature. The earlier treatment yielded biased parameter estimates. An asymptotically unbiased estimator of the autoregressive parameters is obtained as the solution of a modified set of Yule-Walker equations. The asymptotic estimator covariance matrix behaves like a least-squares parameter estimate of an observation set with unknown error covariances. The estimators are also shown to be unbiased in the presence of additive independent observation noise of arbitrary finite correlation time. An example illustrates the performance of the estimating procedures.     

自校正解耦信息融合Wiener状态估值器

对含未知噪声方差阵的多传感器系统,用现代时间序列分析方法.基于滑动平均(MA)新息模型的在线辨识和求解相关函数矩阵方程组,可得到估计噪声方差阵估值器,进而在按分量标量加权线性最小方差最优信息融合则下,提出了自校正解耦信息融合Wiener状态估值器.它的精度比每个局部自校正Wiener状态估值器精度高.它实现了状态分量的解耦局部Wiener估值器和解耦融合Wiener估值器.证明了它的收敛性,即若MA新息模型参数估计是一致的,则它将收敛于噪声统计已知时的最优解耦信息融合Wiener状态估值器,因而它具有渐近最优性.一个带3传感器的目标跟踪系统的仿真例子说明了其有效性.     

Comparison of different pseudo-linear estimators for vision-based target motion estimation

Vision-based target motion estimation based Kalman filtering or least-squares estimators is an important problem in many tasks such as vision-based swarming or vision-based target pursuit. In this paper, we focus on a problem that is very specific yet we believe important. That is, from the vision measurements, we can formulate various measurements. Which and how the measurements should be used? These problems are very fundamental, but we notice that practitioners usually do not pay special attention to them and often make mistakes. Motivated by this, we formulate three pseudo-linear measurements based on the bearing and angle measurements, which are standard vision measurements that can be obtained. Different estimators based on Kalman filtering and least-squares estimation are established and compared based on numerical experiments. It is revealed that correctly analyzing the covariance noises is critical for the Kalman filtering-based estimators. When the variance of the original measurement noise is unknown, the pseudo-linear least-squares estimator that has the smallest magnitude of the transformed noise can be a good choice.     

One-step minimum Hellinger distance estimation

It is well known now that the minimum Hellinger distance estimation approach introduced by Beran (Beran, R., 1977. Minimum Hellinger distance estimators for parametric models. Ann. Statist. 5, 445-463) produces estimators that achieve efficiency at the model density and simultaneously have excellent robustness properties. However, computational difficulties and algorithmic convergence problems associated with this method have hampered its application in practice, particularly when the method is applied to models with high-dimensional parameter spaces. A one-step minimum Hellinger distance (MHD) procedure is investigated in this paper to overcome computational drawbacks of the fully iterative MHD method. The idea is to start with an initial estimator, and then iterate the Newton-Raphson equation once related to the Hellinger distance. The resulting estimator can be considered a one-step MHD estimator. We show that the proposed one-step MHD estimator has the same asymptotic behavior as the MHD estimator, as long as the initial estimators are reasonably good. Furthermore, our theoretical and numerical studies also demonstrate that the proposed one-step MHD estimator also retains excellent robustness properties of the MHD estimators. A real data example is analyzed as well.     

On the performance of the flexible maximum entropy distributions within partially adaptive estimation

The partially adaptive estimation based on the assumed error distribution has emerged as a popular approach for estimating a regression model with non-normal errors. In this approach, if the assumed distribution is flexible enough to accommodate the shape of the true underlying error distribution, the efficiency of the partially adaptive estimator is expected to be close to the efficiency of the maximum likelihood estimator based on knowledge of the true error distribution. In this context, the maximum entropy distributions have attracted interest since such distributions have a very flexible functional form and nest most of the statistical distributions. Therefore, several flexible MaxEnt distributions under certain moment constraints are determined to use within the partially adaptive estimation procedure and their performances are evaluated relative to well-known estimators. The simulation results indicate that the determined partially adaptive estimators perform well for non-normal error distributions. In particular, some can be useful in dealing with small sample sizes. In addition, various linear regression applications with non-normal errors are provided.     

A variational model for normal computation of point clouds

In this paper we present a novel model for computing the oriented normal field on a point cloud. Differently from previous two-stage approaches, our method integrates the unoriented normal estimation and the consistent normal orientation into one variational framework. The normal field with consistent orientation is obtained by minimizing a combination of the Dirichlet energy and the coupled-orthogonality deviation, which controls the normals perpendicular to and continuously varying on the underlying shape. The variational model leads to solving an eigenvalue problem. If unoriented normal field is provided, the model can be modified for consistent normal orientation. We also present experiments which demonstrate that our estimates of oriented normal vectors are accurate for smooth point clouds, and robust in the presence of noise, and reliable for surfaces with sharp features, e.g., corners, ridges, close-by sheets and thin structures.     

Stereoscopic Segmentation

We cast the problem of multiframe stereo reconstruction of a smooth shape as the global region segmentation of a collection of images of the scene. Dually, the problem of segmenting multiple calibrated images of an object becomes that of estimating the solid shape that gives rise to such images. We assume that the radiance of the scene results in piecewise homogeneous image statistics. This simplifying assumption covers Lambertian scenes with constant albedo as well as fine homogeneous textures, which are known challenges to stereo algorithms based on local correspondence. We pose the segmentation problem within a variational framework, and use fast level set methods to find the optimal solution numerically. Our algorithm does not work in the presence of strong photometric features, where traditional reconstruction algorithms do. It enjoys significant robustness to noise under the assumptions it is designed for.     

Robust, reduced-order, nonstrictly proper state estimation via the optimal projection equations with Petersen—Hollot bounds

A state-estimation design problem involving parametric plant uncertainties is considered. An error bound suggested by recent work of Petersen and Hollot is utilized for guaranteeing robust estimation. Necessary conditions which generalize the optimal projection equations for reduced-order state estimation are used to characterize the estimator which minimizes the error bound. The design equations thus effectively serve as sufficient conditions for synthesizing robust estimators. An additional feature is the presence of a static estimation gain in conjunction with the dynamic (Kalman) estimator, i. e., a nonstrictly proper estimator.     

Harmonic point cloud orientation

In this work we propose a new method for estimating the normal orientation of unorganized point clouds. Consistent assignment of normal orientation is a challenging task in the presence of sharp features, nearby surface sheets, noise, undersampling, and missing data. Existing approaches, which consider local geometric properties often fail when operating on such point clouds as local neighborhood measures inherently face issues of robustness. Our approach circumvents these issues by orienting normals based on globally smooth functions defined on point clouds with measures that depend only on single points. More specifically, we consider harmonic functions, or functions which lie in the kernel of the point cloud Laplace-Beltrami operator. Each harmonic function in the set is used to define a gradient field over the point cloud. The problem of normal orientation is then cast as an assignment of cross-product ordering between gradient fields. Global smoothness ensures a highly consistent orientation, rendering our method extremely robust in the presence of imperfect point clouds.     

On the stability of some high-resolution beamforming methods

Orthogonal beamforming is the name of certain high-resolution methods for estimating the spectra of a wave field received by an array of sensors. The methods use the eigenvalues and eigenvectors of the spectral matrix of the sensor outputs. The problem is to predict the behavior of such methods when only an estimate of the matrix is known. The sensor outputs may consist of sensor noise, ambient noise and noise from a finite set of discrete sources. The properties of the eigensystem of the spectral matrix in the case of weak ambient noise motivate the methods of orthogonal beamforming, for example Pisarenko's nonlinear peak estimates and the projection estimates of Owsley. If the spectral matrix is estimated by one of the classical methods, some asymptotic distributional properties of the matrix estimate and its eigensystem are well known. They can be used to determine asymptotic expressions, e.g. for the first and second moments of the peak estimators and to approximate the distributions. The parameters, however, cannot be calculated in applications, since the eigensystem of the exact spectral matrix is required. Therefore, we have developed bounds for the deviation of the peak estimates which only use weak knowledge about the matrix. We have applied some results on perturbations of hermitian operators. The asymptotic behavior of the bounds for the projection estimator is investigated and possibilities for their estimation are indicated. Finally, we report on extensive simulations with random matrices to evaluate the new bounds. As a result, we have found that the projection estimator behaves stably and there are tight bounds if the eigenvalues of interest are sufficiently separated from the rest.     

Robust H∞ Filtering and Deconvolution for Continuous Time Delay Systems Based on Game‐Theoretic Approach

Many dynamical systems involve not only process and measurement noise signals but also parameter uncertainty and unknown input signals. This paper aims to estimate the state and unknown input for linear continuous time‐varying systems subject to time delay in state, norm‐bounded parameter uncertainty, and a known input. Such a problem is reformulated into a two‐player differential game whose saddle point solution gives rise to one sufficient solvable condition for the estimation problem. The possible optimal estimators are obtained by solving the two coupled Riccati differential equations. We demonstrate, through two examples, how the proposed estimator is valid for estimating state and unknown input.