non-Gaussian Case Research Articles

Application of Machine Learning Tools for Long-Term Diagnostic Feature Data Segmentation

In this paper, a novel method for long-term data segmentation in the context of machine health prognosis is presented. The purpose of the method is to find borders between three data segments. It is assumed that each segment contains the data that represent different statistical properties, that is, a different model. It is proposed to use a moving window approach, statistical parametrization of the data in the window, and simple clustering techniques. Moreover, it is found that features are highly correlated, so principal component analysis is exploited. We find that the probability density function of the first principal component may be sufficient to find borders between classes. We consider two cases of data distributions, Gaussian and α-stable, belonging to the class of non-Gaussian heavy-tailed distributions. It is shown that for random components with Gaussian distribution, the proposed methodology is very effective, while for the non-Gaussian case, both features and the concept of moving window should be re-considered. Finally, the procedure is tested for real data sets. The results provided here may be helpful in understanding some specific cases of machine health prognosis in the presence of non-Gaussian noise. The proposed approach is model free, and thus it is universal. The methodology can be applied for any long-term data where segmentation is crucial for the data processing.

An optimal control method to compute the most likely transition path for stochastic dynamical systems with jumps.

Many complex real world phenomena exhibit abrupt, intermittent, or jumping behaviors, which are more suitable to be described by stochastic differential equations under non-Gaussian Lévy noise. Among these complex phenomena, the most likely transition paths between metastable states are important since these rare events may have a high impact in certain scenarios. Based on the large deviation principle, the most likely transition path could be treated as the minimizer of the rate function upon paths that connect two points. One of the challenges to calculate the most likely transition path for stochastic dynamical systems under non-Gaussian Lévy noise is that the associated rate function cannot be explicitly expressed by paths. For this reason, we formulate an optimal control problem to obtain the optimal state as the most likely transition path. We then develop a neural network method to solve this issue. Several experiments are investigated for both Gaussian and non-Gaussian cases.

Stein Particle Filter for Nonlinear, Non-Gaussian State Estimation

Estimation of a dynamical system’s latent state subject to sensor noise and model inaccuracies remains a critical yet difficult problem in robotics. While Kalman filters provide the optimal solution in the least squared sense for linear and Gaussian noise problems, the general nonlinear and non-Gaussian noise case is significantly more complicated, typically relying on sampling strategies that are limited to low-dimensional state spaces. In this letter, we devise a general inference procedure for filtering of nonlinear, non-Gaussian dynamical systems that exploits the differentiability of both the update and prediction models to scale to higher dimensional spaces. Our method, Stein particle filter, can be seen as a deterministic flow of particles, embedded in a reproducing kernel Hilbert space, from an initial state to the desirable posterior. The particles evolve jointly to conform to a posterior approximation while interacting with each other through a repulsive force. We evaluate the method in simulation and in complex localization tasks while comparing it to sequential Monte Carlo solutions.

Functional principal component analysis estimator for non-Gaussian data

Functional principal component analysis (FPCA) could become invalid when data involve non-Gaussian features. Therefore, we aim to develop a general FPCA method that can adapt to such non-Gaussian cases. A Kendall's τ function, which possesses identical eigenfunctions as covariance function, is constructed. The particular formulation of Kendall's τ function makes it less sensitive to data distribution. We further apply it to the estimation of FPCA and study the corresponding asymptotic consistency. Moreover, the effectiveness of the proposed method is demonstrated through a comprehensive simulation study and an application to the physical activity data that collected by a wearable accelerometer monitor.

Mixture Correntropy-Based Kernel Extreme Learning Machines.

Kernel-based extreme learning machine (KELM), as a natural extension of ELM to kernel learning, has achieved outstanding performance in addressing various regression and classification problems. Compared with the basic ELM, KELM has a better generalization ability owing to no needs of the number of hidden nodes given beforehand and random projection mechanism. Since KELM is derived under the minimum mean square error (MMSE) criterion for the Gaussian assumption of noise, its performance may deteriorate under the non-Gaussian cases, seriously. To improve the robustness of KELM, this article proposes a mixture correntropy-based KELM (MC-KELM), which adopts the recently proposed maximum mixture correntropy criterion as the optimization criterion, instead of using the MMSE criterion. In addition, an online sequential version of MC-KELM (MCOS-KELM) is developed to deal with the case that the data arrive sequentially (one-by-one or chunk-by-chunk). Experimental results on regression and classification data sets are reported to validate the performance superiorities of the new methods.

Recursive minimum kernel risk sensitive loss algorithm with adaptive gain factor for robust power system s estimation

The development of effective harmonics estimation method plays a crucial role for safe operation of power system. Recently several existing parametric estimation approaches, however, are mainly developed and discussed under the assumption of Gaussian noise which is always unsuitable in real power systems. This work is mainly concerned with the robust parametric estimation methods. The kernel risk sensitive loss (KRSL) shows an advantage in that its performance surface is flat around the optimal solution while sharp far away from the optimal solution and it may exhibit outstanding performance in terms of faster convergence rate and higher estimation accuracy under non-Gaussian noise cases. Therefore, a minimum KRSL (MKRSL) algorithm with sinusoidal signal is designed, and we further develop a novel recursive MKRSL (RMKRSL) algorithm for harmonics estimation in non-Gaussian noise cases. In addition, an adaptive gain factor is introduced into the gain matrix of RMKRSL algorithm to enhance its tracking capability. Some simulations by using the signal generated from the given models and IEEE-1159-PQE databases are conducted to illustrate the performance of the proposed methods under different conditions.

Diffusion Generalized MCC with a Variable Center Algorithm for Robust Distributed Estimation

Classical adaptive filtering algorithms with a diffusion strategy under the mean square error (MSE) criterion can face difficulties in distributed estimation (DE) over networks in a complex noise environment, such as non-zero mean non-Gaussian noise, with the object of ensuring a robust performance. In order to overcome such limitations, this paper proposes a novel robust diffusion adaptive filtering algorithm, which is developed by using a variable center generalized maximum Correntropy criterion (GMCC-VC). Generalized Correntropy with a variable center is first defined by introducing a non-zero center to the original generalized Correntropy, which can be used as robust cost function, called GMCC-VC, for adaptive filtering algorithms. In order to improve the robustness of the traditional MSE-based DE algorithms, the GMCC-VC is used in a diffusion adaptive filter to design a novel robust DE method with the adapt-then-combine strategy. This can achieve outstanding steady-state performance under non-Gaussian noise environments because the GMCC-VC can match the distribution of the noise with that of non-zero mean non-Gaussian noise. The simulation results for distributed estimation under non-zero mean non-Gaussian noise cases demonstrate that the proposed diffusion GMCC-VC approach produces a more robustness and stable performance than some other comparable DE methods.

Competition on Spatial Statistics for Large Datasets

As spatial datasets are becoming increasingly large and unwieldy, exact inference on spatial models becomes computationally prohibitive. Various approximation methods have been proposed to reduce the computational burden. Although comprehensive reviews on these approximation methods exist, comparisons of their performances are limited to small and medium sizes of datasets for a few selected methods. To achieve a comprehensive comparison comprising as many methods as possible, we organized the Competition on Spatial Statistics for Large Datasets. This competition had the following novel features: (1) we generated synthetic datasets with the ExaGeoStat software so that the number of generated realizations ranged from 100 thousand to 1 million; (2) we systematically designed the data-generating models to represent spatial processes with a wide range of statistical properties for both Gaussian and non-Gaussian cases; (3) the competition tasks included both estimation and prediction, and the results were assessed by multiple criteria; and (4) we have made all the datasets and competition results publicly available to serve as a benchmark for other approximation methods. In this paper, we disclose all the competition details and results along with some analysis of the competition outcomes.

Extreme estimation of wind pressure with unimodal and bimodal probability density function characteristics: A maximum entropy model based on fractional moments

The complicated probability density function (PDF) characteristics of wind pressure, including its highly skewed, highly leptokurtic, and bimodal characteristics induced by sophisticated architecture, call for improved extreme-estimation models. Most existing methods are based on a unimodal PDF's underlying assumption and fail in highly non-Gaussian cases. Owing to the complete probabilistic information used in cumulative distribution function (CDF)-mapping methods, such methods have the potential to deal with complicated PDF-type wind pressures. To yield better curve-fitting of the parent distribution, which is the critical step of CDF-mapping, a maximum entropy model based on fractional moments is combined with CDF-mapping in this work. By using the Legendre-Gauss quadrature rule, the computational efficiency of fitting parent distribution is improved. The model's performance is benchmarked against typical long-term wind pressure data obtained from wind tunnel tests conducted at a long-span airport terminal model. By considering the captured probability properties, four typical taps are selected to provide empirical wind-pressure extreme in the 57% fractile for detailed model assessment. The confidence intervals of the estimated results and errors for all taps are also calculated. Compared with existing polynomial-fitting- and CDF-mapping-based translation methods, the extreme estimated by the proposed method are shown to be more robust and stable. • A novel approach was proposed for extreme estimation of wind pressures with unimodal and bimodal cases. • The Legendre-Gauss quadrature rule was adapted for improving the model computational efficiency. • The model performance was verified with long-term measured data from the wind tunnel test of a long-span building. • A detailed comparison against existing translation-based methods was conducted.

SwarmCCO: Probabilistic Reactive Collision Avoidance for Quadrotor Swarms Under Uncertainty

We present decentralized collision avoidance algorithms for quadrotor swarms operating under uncertain state estimation. Our approach exploits the differential flatness property and feedforward linearization to approximate the quadrotor dynamics and performs reciprocal collision avoidance. We account for the uncertainty in position and velocity by formulating the collision constraints as chance constraints, which describe a set of velocities that avoid collisions with a specified confidence level. We present two different methods for formulating and solving the chance constraints: our first method assumes a Gaussian noise distribution. Our second method is its extension to the non-Gaussian case by using a Gaussian Mixture Model (GMM). We reformulate the linear chance constraints into equivalent deterministic constraints, which are used with an MPC framework to compute a local collision-free trajectory for each quadrotor. We evaluate the proposed algorithm in simulations on benchmark scenarios and highlight its benefits over prior methods. We observe that both the Gaussian and non-Gaussian methods provide improved collision avoidance performance over the deterministic method. On average, the Gaussian method requires ~ 5 ms to compute a local collision-free trajectory, while our non-Gaussian method is computationally more expensive and requires ~ 9 ms on average in scenarios with 4 agents.

Addressing indirect frequency coupling via partial generalized coherence

Distinguishing between direct and indirect frequency coupling is an important aspect of functional connectivity analyses because this distinction can determine if two brain regions are directly connected. Although partial coherence quantifies partial frequency coupling in the linear Gaussian case, we introduce a general framework that can address even the nonlinear and non-Gaussian case. Our technique, partial generalized coherence (PGC), expands prior work by allowing pairwise frequency coupling analyses to be conditioned on other processes, enabling model-free partial frequency coupling results. By taking advantage of recent advances in conditional mutual information estimation, we are able to implement our technique in a way that scales well with dimensionality, making it possible to condition on many processes and produce a partial frequency coupling graph. We analyzed both linear Gaussian and nonlinear simulated networks. We then performed PGC analysis of calcium recordings from mouse olfactory bulb glomeruli under anesthesia and quantified the dominant influence of breathing-related activity on the pairwise relationships between glomeruli for breathing-related frequencies. Overall, we introduce a technique capable of eliminating indirect frequency coupling in a model-free way, empowering future research to correct for potentially misleading frequency interactions in functional connectivity analyses.

A surrogate-based approach to nonlinear, non-Gaussian joint state-parameter data assimilation

<p style='text-indent:20px;'>Many recent advances in sequential assimilation of data into nonlinear high-dimensional models are modifications to particle filters which employ efficient searches of a high-dimensional state space. In this work, we present a complementary strategy that combines statistical emulators and particle filters. The emulators are used to learn and offer a computationally cheap approximation to the forward dynamic mapping. This emulator-particle filter (Emu-PF) approach requires a modest number of forward-model runs, but yields well-resolved posterior distributions even in non-Gaussian cases. We explore several modifications to the Emu-PF that utilize mechanisms for dimension reduction to efficiently fit the statistical emulator, and present a series of simulation experiments on an atypical Lorenz-96 system to demonstrate their performance. We conclude with a discussion on how the Emu-PF can be paired with modern particle filtering algorithms.</p>

Minimum Error Entropy Based EKF for GPS Code Tracking Loop

This paper investigates the minimum error entropy based extended Kalman filter (MEEKF) for multipath parameter estimation of the Global Positioning System (GPS). The extended Kalman filter (EKF) is designed to give a preliminary estimation of the state. The scheme is designed by introducing an additional term, which is tuned according to the higher order moment of the estimation error. The minimum error entropy criterion is introduced for updating the entropy of the innovation at each time step. According to the stochastic information gradient method, an optimal filer gain matrix is obtained. The mean square error criterion is limited to the assumption of linearity and Gaussianity. However, non-Gaussian noise is often encountered in many practical environments and their performances degrade dramatically in non-Gaussian cases. Most of the existing multipath estimation algorithms are usually designed for Gaussian noise. The I (in-phase) and Q (quadrature) accumulator outputs from the GPS correlators are used as the observational measurements of the EKF to estimate the multipath parameters such as amplitude, code delay, phase, and carrier Doppler. One reasonable way to obtain an optimal estimation is based on the minimum error entropy criterion. The MEEKF algorithm provides better estimation accuracy since the error entropy involved can characterize all the randomness of the residual. Performance assessment is presented to evaluate the effectivity of the system designs for GPS code tracking loop with multipath parameter estimation using the minimum error entropy based extended Kalman filter.

Mean field approach to stochastic control with partial information

In our present article, we follow our way of developing mean field type control theory in our earlier works [Bensoussanet al., Mean Field Games and Mean Field Type Control Theory.Springer, New York (2013)], by first introducing the Bellman and then master equations, the system of Hamilton-Jacobi-Bellman (HJB) and Fokker-Planck (FP) equations, and then tackling them by looking for the semi-explicit solution for the linear quadratic case, especially with an arbitrary initial distribution; such a problem, being left open for long, has not been specifically dealt with in the earlier literature, such as Bensoussan [Stochastic Control of Partially Observable Systems. Cambridge University Press, (1992)] and Nisio [Stochastic control theory: Dynamic programming principle. Springer (2014)], which only tackled the linear quadratic setting with Gaussian initial distributions. Thanks to the effective mean-field theory, we propose a solution to this long standing problem of the general non-Gaussian case. Besides, our problem considered here can be reduced to the model in Bandiniet al.[Stochastic Process. Appl.129(2019) 674–711], which is fundamentally different from our present proposed framework.

Correlators in the Gaussian and chiral supereigenvalue models in the Neveu-Schwarz sector

We analyze the Gaussian and chiral supereigenvalue models in the Neveu-Schwarz sector. We show that their partition functions can be expressed as the infinite sums of the homogeneous operators acting on the elementary functions. In spite of the fact that the usual W-representations of these matrix models can not be provided here, we can still derive the compact expressions of the correlators in these two supereigenvalue models. Furthermore, the non-Gaussian (chiral) cases are also discussed.

Steady State Mean Square Analysis of Standard Maximum Versoria Criterion Based Adaptive Algorithm

The Maximum Versoria Criterion (MVC) based adaptive algorithm has recently gained the attention of researchers because of its robustness against impulsive interference and reduced complexity. In this brief, the steady-state Excess Mean Square Error (EMSE) analysis of the standard MVC algorithm is done using the energy conservation relation. For the Gaussian noise case, quadratic equation that can provide the exact value of theoretical steady state EMSE is obtained without any assumptions other than the conventional ones and the Price's theorem. For the non-Gaussian case, an approximate solution using Taylor's expansion is derived. Experimental results validate the obtained theoretical findings.

Optimal lower bounds on hitting probabilities for non-linear systems of stochastic fractional heat equations

We consider a system of d non-linear stochastic fractional heat equations in spatial dimension 1 driven by multiplicative d-dimensional space–time white noise. We establish a sharp Gaussian-type upper bound on the two-point probability density function of (u(s,y),u(t,x)). From this result, we deduce optimal lower bounds on hitting probabilities of the process {u(t,x):(t,x)∈[0,∞[×R} in the non-Gaussian case, in terms of Newtonian capacity, which is as sharp as that in the Gaussian case. This also improves the result in Dalang et al. (2009) for systems of classical stochastic heat equations. We also establish upper bounds on hitting probabilities of the solution in terms of Hausdorff measure.

Mean exit time and escape probability for the stochastic logistic growth model with multiplicative α-stable Lévy noise

In this paper, we formulate a stochastic logistic fish growth model driven by both white noise and non-Gaussian noise. We focus our study on the mean time to extinction, escape probability to measure the noise-induced extinction probability and the Fokker–Planck equation for fish population [Formula: see text]. In the Gaussian case, these quantities satisfy local partial differential equations while in the non-Gaussian case, they satisfy nonlocal partial differential equations. Following a discussion of existence, uniqueness and stability, we calculate numerical approximations of the solutions of those equations. For each noise model we then compare the behaviors of the mean time to extinction and the solution of the Fokker–Planck equation as growth rate [Formula: see text], carrying capacity [Formula: see text], intensity of Gaussian noise [Formula: see text], noise intensity [Formula: see text] and stability index [Formula: see text] vary. The MET from the interval [Formula: see text] at the right boundary is finite if [Formula: see text]. For [Formula: see text], the MET from [Formula: see text] at this boundary is infinite. A larger stability index [Formula: see text] is less likely leading to the extinction of the fish population.

Learning Causal Structures Based on Divide and Conquer

This article addresses two important issues of causal inference in the high-dimensional situation. One is how to reduce redundant conditional independence (CI) tests, which heavily impact the efficiency and accuracy of existing constraint-based methods. Another is how to construct the true causal graph from a set of Markov equivalence classes returned by these methods. For the first issue, we design a recursive decomposition approach where the original data (a set of variables) are first decomposed into two small subsets, each of which is then recursively decomposed into two smaller subsets until none of these subsets can be decomposed further. Redundant CI tests can be reduced by inferring causalities from these subsets. The advantage of this decomposition scheme lies in two aspects: 1) it requires only low-order CI tests and 2) it does not violate d -separation. The complete causality can be reconstructed by merging all the partial results of the subsets. For the second issue, we employ regression-based CI tests to check CIs in linear non-Gaussian additive noise cases, which can identify more causal directions by [Formula: see text] (or [Formula: see text]). Consequently, causal direction learning is no longer limited by the number of returned V -structures and consistent propagation. Extensive experiments show that the proposed method can not only substantially reduce redundant CI tests but also effectively distinguish the equivalence classes.

Generation of strongly non-Gaussian stochastic processes by iterative scheme upgrading phase and amplitude contents

Random excitations, such as wind velocity, always exhibit non-Gaussian features. Sample realisations of stochastic processes satisfying given features should be generated, in order to perform the dynamical analysis of structures under stochastic loads based on the Monte Carlo simulation. In this paper, an efficient method is proposed to generate stationary non-Gaussian stochastic processes. It involves an iterative scheme that produces a class of sample processes satisfying the following conditions. (1) The marginal cumulative distribution function of each sample process is perfectly identical to the prescribed one. (2) The ensemble-averaged power spectral density function of these non-Gaussian sample processes is as close to the prescribed target as possible. In this iterative scheme, the underlying processes are generated by means of the spectral representation method that recombines the upgraded power spectral density function with the phase contents of the new non-Gaussian processes in the latest iteration. Numerical examples are provided to demonstrate the capabilities of the proposed approach for four typical non-Gaussian distributions, some of which deviate significantly from the Gaussian distribution. It is found that the estimated power spectral density functions of non-Gaussian processes are close to the target ones, even for the extremely non-Gaussian case. Furthermore, the capability of the proposed method is compared to two other methods. The results show that the proposed method performs well with convergence speed, accuracy, and random errors of power spectral density functions.