Application of the maximum likelihood principle to separate exponential terms in T2 relaxation of nuclear magnetic resonance

The method of maximum likelihood has been implemented for the estimation of multiple exponential components of T2 decay curves in spin echo NMR measurements on biologic tissues. Each Each component contributes an exponential term described by two parameters (initial amplitude and T2) to the T2 decay curve. The maximum likelihood method estimates the parameters and their standard errors for all terms simultaneously, avoiding the subjectivity inherent in methods such as graphical peeling. In the model used, it was assumed that water protons are compartmentalized and that the measured spin echo signals from the protons undergoing relaxation obey the Poisson distribution. A system of non-linear equations was derived and solved iteratively for the values of the exponential parameters which maximize the likelihood of obtaining the observed data under these assumptions. The approach was implemented for bi- and tri-exponential models on a MicroVAX II computer (Digital Equipment Corporation, Maynard, MA). Simulations of bi- and tri-exponential data, with and without system noise, were analyzed to assess the accuracy and reproducibility of the method. A subset of the simulations was repeated with non-linear least squares techniques and was compared to the results obtained with maximum likelihood. Rabbit muscle and gerbil brain samples were measured and analyzed with the maximum likelihood method. The simulations showed that within specific limits on relative sizes and relaxation rates of components, these parameters can be estimated with errors less than 5%. The comparison to non-linear least squares analysis showed that the maximum likelihood method is generally superior in estimating the parameters in difficult cases. The results from tissue measurements demonstrate that the method is effective even in cases where graphical peeling would clearly not yield reliable results.     

Bayesian and Frequentist Inferences on a Type I Half-Logistic Odd Weibull Generator with Applications in Engineering

In this article, we have proposed a new generalization of the odd Weibull-G family by consolidating two notable families of distributions. We have derived various mathematical properties of the proposed family, including quantile function, skewness, kurtosis, moments, incomplete moments, mean deviation, Bonferroni and Lorenz curves, probability weighted moments, moments of (reversed) residual lifetime, entropy and order statistics. After producing the general class, two of the corresponding parametric statistical models are outlined. The hazard rate function of the sub-models can take a variety of shapes such as increasing, decreasing, unimodal, and Bathtub shaped, for different values of the parameters. Furthermore, the sub-models of the introduced family are also capable of modelling symmetric and skewed data. The parameter estimation of the special models are discussed by numerous methods, namely, the maximum likelihood, simple least squares, weighted least squares, Cramér-von Mises, and Bayesian estimation. Under the Bayesian framework, we have used informative and non-informative priors to obtain Bayes estimates of unknown parameters with the squared error and generalized entropy loss functions. An extensive Monte Carlo simulation is conducted to assess the effectiveness of these estimation techniques. The applicability of two sub-models of the proposed family is illustrated by means of two real data sets.     

The Anderson-Darling test of fit for the power-law distribution from left-censored samples

Maximum likelihood estimation and a test of fit based on the Anderson-Darling statistic are presented for the case of the power-law distribution when the parameters are estimated from a left-censored sample. Expressions for the maximum likelihood estimators and tables of asymptotic percentage points for the A² statistic are given. The technique is illustrated for data from the Dow Jones Industrial Average index, an example of high theoretical and practical importance in Econophysics, Finance, Physics, Biology and, in general, in other related sciences such as Complexity Sciences.     

The Truncated Burr X-G Family of Distributions: Properties and Applications to Actuarial and Financial Data

In this article, the “truncated-composed” scheme was applied to the Burr X distribution to motivate a new family of univariate continuous-type distributions, called the truncated Burr X generated family. It is mathematically simple and provides more modeling freedom for any parental distribution. Additional functionality is conferred on the probability density and hazard rate functions, improving their peak, asymmetry, tail, and flatness levels. These characteristics are represented analytically and graphically with three special distributions of the family derived from the exponential, Rayleigh, and Lindley distributions. Subsequently, we conducted asymptotic, first-order stochastic dominance, series expansion, Tsallis entropy, and moment studies. Useful risk measures were also investigated. The remainder of the study was devoted to the statistical use of the associated models. In particular, we developed an adapted maximum likelihood methodology aiming to efficiently estimate the model parameters. The special distribution extending the exponential distribution was applied as a statistical model to fit two sets of actuarial and financial data. It performed better than a wide variety of selected competing non-nested models. Numerical applications for risk measures are also given.     

Information-Corrected Estimation: A Generalization Error Reducing Parameter Estimation Method

Modern computational models in supervised machine learning are often highly parameterized universal approximators. As such, the value of the parameters is unimportant, and only the out of sample performance is considered. On the other hand much of the literature on model estimation assumes that the parameters themselves have intrinsic value, and thus is concerned with bias and variance of parameter estimates, which may not have any simple relationship to out of sample model performance. Therefore, within supervised machine learning, heavy use is made of ridge regression (i.e., L2 regularization), which requires the the estimation of hyperparameters and can be rendered ineffective by certain model parameterizations. We introduce an objective function which we refer to as Information-Corrected Estimation (ICE) that reduces KL divergence based generalization error for supervised machine learning. ICE attempts to directly maximize a corrected likelihood function as an estimator of the KL divergence. Such an approach is proven, theoretically, to be effective for a wide class of models, with only mild regularity restrictions. Under finite sample sizes, this corrected estimation procedure is shown experimentally to lead to significant reduction in generalization error compared to maximum likelihood estimation and L2 regularization.     

Statistical Inference of Truncated Normal Distribution Based on the Generalized Progressive Hybrid Censoring

In this paper, the parameter estimation problem of a truncated normal distribution is discussed based on the generalized progressive hybrid censored data. The desired maximum likelihood estimates of unknown quantities are firstly derived through the Newton–Raphson algorithm and the expectation maximization algorithm. Based on the asymptotic normality of the maximum likelihood estimators, we develop the asymptotic confidence intervals. The percentile bootstrap method is also employed in the case of the small sample size. Further, the Bayes estimates are evaluated under various loss functions like squared error, general entropy, and linex loss functions. Tierney and Kadane approximation, as well as the importance sampling approach, is applied to obtain the Bayesian estimates under proper prior distributions. The associated Bayesian credible intervals are constructed in the meantime. Extensive numerical simulations are implemented to compare the performance of different estimation methods. Finally, an authentic example is analyzed to illustrate the inference approaches.     

Efficient image segmentation method based on resolution and region information fusion

A generalized multiresolution likelihood ratio (GMLR), which can increase the distinction between differentsignals by fusing their more features, is defined. Multiresolution representation of image characterizes in-herent structure of image well, and the GMLR combines each resolution image features with correspondingregion features. A spatially variant mixture multiscale autoregressive prediction (SVMMARP) model isproposed to estimate the parameters of GMLR based on maximum likelihood estimation via expectationmaximization (EM) algorithm. In the parameter estimation, bootstrap sampling technique is employed.Experimental results demonstrate that the algorithm performs fairly well.     

Monitoring Parameter Change for Time Series Models of Counts Based on Minimum Density Power Divergence Estimator

In this study, we consider an online monitoring procedure to detect a parameter change for integer-valued generalized autoregressive heteroscedastic (INGARCH) models whose conditional density of present observations over past information follows one parameter exponential family distributions. For this purpose, we use the cumulative sum (CUSUM) of score functions deduced from the objective functions, constructed for the minimum power divergence estimator (MDPDE) that includes the maximum likelihood estimator (MLE), to diminish the influence of outliers. It is well-known that compared to the MLE, the MDPDE is robust against outliers with little loss of efficiency. This robustness property is properly inherited by the proposed monitoring procedure. A simulation study and real data analysis are conducted to affirm the validity of our method.     

State and parameter estimation in stochastic dynamical models

This paper derives generalized maximum likelihood estimates of state and model parameters of a stochastic dynamical model. In contrast to previous studies, the change in background distribution due to changes in model parameters is taken into account. An ensemble approach to solving the maximum likelihood estimates is proposed. An exact solution for the ensemble update based on a square root Kalman Filter is derived. This solution involves a two step procedure in which an ensemble is first produced by a standard ensemble Kalman Filter, and then “corrected” to account for parameter estimation, thereby allowing a user to take advantage of an existing ensemble filter. The solution is illustrated with simple, low-dimensional stochastic dynamical models and shown to work well and outperform augmentation methods for estimating stochastic parameters.     

Motile parameters of cell migration in anisotropic environment derived by speed power spectrum fitting with double exponential decay

Cell migration through anisotropic microenvironment is critical to a wide variety of physiological and pathological processes.However,adequate analytical tools to derive motile parameters to characterize the anisotropic migration are lacking.Here,we proposed a method to obtain the four motile parameters of migration cells based on the anisotropic persistent random walk model which is described by two persistence times and two migration speeds at perpendicular directions.The key process is to calculate the velocity power spectra of cell migration along intrinsically perpendicular directions respectively,then to apply maximum likelihood estimation to derive the motile parameters from the power spectra fitting with double exponential decay.The simulation results show that the averaged persistence times and the corrected migration speeds can be good estimations for motile parameters of cell migration.     

Variational Bayesian identification and prediction of stochastic nonlinear dynamic causal models

In this paper, we describe a general variational Bayesian approach for approximate inference on nonlinear stochastic dynamic models. This scheme extends established approximate inference on hidden-states to cover: (i) nonlinear evolution and observation functions, (ii) unknown parameters and (precision) hyperparameters and (iii) model comparison and prediction under uncertainty. Model identification or inversion entails the estimation of the marginal likelihood or evidence of a model. This difficult integration problem can be finessed by optimising a free-energy bound on the evidence using results from variational calculus. This yields a deterministic update scheme that optimises an approximation to the posterior density on the unknown model variables. We derive such a variational Bayesian scheme in the context of nonlinear stochastic dynamic hierarchical models, for both model identification and time-series prediction. The computational complexity of the scheme is comparable to that of an extended Kalman filter, which is critical when inverting high dimensional models or long time-series. Using Monte-Carlo simulations, we assess the estimation efficiency of this variational Bayesian approach using three stochastic variants of chaotic dynamic systems. We also demonstrate the model comparison capabilities of the method, its self-consistency and its predictive power.     

期望最大化算法近场被动定位技术研究

本文介绍了一种对近场声源的距离和方位参数的确定性最大似然估计方法。直接的对近场声源参数的最大似然估计产生了复杂的多参数优化问题,我们在实际采样数据(非完全数据)和假设数据(完全数据)等方面重新构建这个问题,最后提出运用期望最大化迭代方法获得最大似然估计。期望最大化算法将观测数据分解,然后对于最优化问题,运用有效的计算措施单独估计每个信号成分的参数。这种算法的应用性和有效性通过一定的仿真得到了验证。     

基于机器视觉的最大似然位姿估计算法

针对现有位姿估计算法对采样数据不做任何的统计假设,缺少评判标准等问题,从信号的概率密度函数出发,推导了基于机器视觉的最大似然位姿估计的一般形式,并证明利用单幅图像时,在各向同性高斯噪声情况下传统迭代算法与最大似然估计等效。推导了位姿估计的克拉美-罗界,给出了位姿估计的方差下限。根据仿真结果可以看出,利用10张图像时,最大似然算法在噪声功率大于5dB的情况下,性能明显优于传统迭代算法,证明适当增加图像数可有效提高估计性能。     

A New Extension of Thinning-Based Integer-Valued Autoregressive Models for Count Data

The thinning operators play an important role in the analysis of integer-valued autoregressive models, and the most widely used is the binomial thinning. Inspired by the theory about extended Pascal triangles, a new thinning operator named extended binomial is introduced, which is a general case of the binomial thinning. Compared to the binomial thinning operator, the extended binomial thinning operator has two parameters and is more flexible in modeling. Based on the proposed operator, a new integer-valued autoregressive model is introduced, which can accurately and flexibly capture the dispersed features of counting time series. Two-step conditional least squares (CLS) estimation is investigated for the innovation-free case and the conditional maximum likelihood estimation is also discussed. We have also obtained the asymptotic property of the two-step CLS estimator. Finally, three overdispersed or underdispersed real data sets are considered to illustrate a superior performance of the proposed model.     

基于最大似然的单通道交叠激光微多普勒信号参数分离估计

利用激光对目标微弱振动进行探测有利于获得明显的微多普勒效应,这为精确估计目标微动特征参数,实现对目标的分类和精细识别提供了可能.但对于多散射点或多目标激光探测,信号为单通道多分量微动混合的形式,而且补偿目标主体运动后,数值上相近的微动参数还会导致信号在时频域存在严重的交叠.为从这类混合信号中精确估计各分量的微动参数,本文提出了基于最大似然框架的参数分离估计方法.利用精细化扫描的奇异值比谱法从混合信号中获得目标微动频率,并得到各分量的幅值比信息.推导了微动参数最大似然估计的解析表达形式,根据激光微多普勒信号的特点从频谱能量分布的角度重新设计了似然函数,解决了传统似然函数在激光微动信号中出现的高度非线性问题,降低了初始化的要求,提高了抗噪性能,并采用马尔可夫-蒙特卡罗方法具体实现了参数的估计.在微动参数得到估计的基础上给出了信号的幅值和初相的估计方法.用本文方法对仿真和实验数据进行处理,得到了接近克拉美罗下界的估计结果,验证了方法的有效性.与传统非参数化估计方法的对比结果体现了所提方法对混合微动参数精确估计上的优势.     

The Lyapunov Exponent of Products of Random 2×2 Matrices Close to the Identity

We study products of arbitrary random real 2×2 matrices that are close to the identity matrix. Using the Iwasawa decomposition of SL(2,?), we identify a continuum regime where the mean values and the covariances of the three Iwasawa parameters are simultaneously small. In this regime, the Lyapunov exponent of the product is shown to assume a scaling form. In the general case, the corresponding scaling function is expressed in terms of Gauss’ hypergeometric function. A number of particular cases are also considered, where the scaling function of the Lyapunov exponent involves other special functions (Airy, Bessel, Whittaker, elliptic). The general solution thus obtained allows us, among other things, to recover in a unified framework many results known previously from exactly solvable models of one-dimensional disordered systems.     

一种基于最大似然的混响时间盲估计方法*

混响是室内声学中的重要现象,在室内设计与音频信号处理中都需要测量或估计混响时间。本文改进了一种基于最大似然估计的混响时间盲估计方法,即采用说话人在房间中自然说话时发出的混响语音信号来估计混响时间的方法。该方法首先确定语音衰减段的最优边界,其次计算该衰减段的两个额外参数,据此筛选出符合条件的语音段,最后将满足条件的语音段采用最大似然估计得到混响时间估计值。在五个不同混响时间条件下的仿真表明,与已有方法相比,改进方法估计的混响时间同真实混响时间的偏差更小,方差更低,估计准确性较高。     

Near optimal performance joint semi-blind channel estimation and data detection techniques for Alamouti coded single-carrier (SC) MIMO communication systems

In this paper, two novel joint semi-blind channel estimation and data detection techniques are proposed and investigated for Alamouti coded single-carrier (SC) multiple-input multiple-output (MIMO) communication system using Rayleigh flat fading channel model. In the first novel semi-blind technique, blind channel estimation can be performed by using singular value decomposition (SVD) of received output autocorrelation matrix and training based channel estimation for orthogonal training symbols can be performed by using orthogonal pilot maximum likelihood (OPML) algorithm. Further using, that semi-blind channel estimate and received output, data detection is performed by using Maximum likelihood (ML) detection. Finally we derived new training symbols from error covariance matrix of estimated data and known orthogonal training symbols, which further applied to OPML algorithm for final channel estimate. In the second novel semi-blind technique, blind channel estimation can be performed by using matrix triangularization based on householder QR decomposition (H-QRD) of received output autocorrelation matrix instead of SVD decomposition. Other steps are same as the first novel technique to calculate data detection and final channel estimation. Simulation results are presented under 2-PSK, 4-PSK, 8-PSK and 16-QAM data modulation schemes using 2 transmitters and different combinations of receiver antennas to investigate the performances of novel techniques compare to conventional whitening rotation (WR) and rotation optimization maximum likelihood (ROML) based semi-blind channel estimation techniques. Result demonstrates that novel techniques outperform others by achieving near optimal performance.     

A frequency domain inversion method applied to propagation models in unconsolidated granular materials

The measurement setup consists of a water filled tank within which a smaller Plexiglas tank containing a known sediment is placed. This provides a Plexiglas-sediment-Plexiglas configuration, which is then used in conjunction with transducers and a pulser-receiver for a transmission and reflection experiment. The aim of this work is to find some physical parameters of the sediment from these measurements. A general viscoelastic model that considers losses due to absorption and dispersion along the propagation of the wave through the sediments is proposed. A rational transfer function is used to model the viscoelastic modulus of the bulk sediments, and comparisons with constant Q viscoelastic modelling and Biot modelling are performed. The estimation of the model parameters is elaborated using a maximum likelihood estimator in the frequency domain.