Stochastic approximation learning for mixtures of multivariate elliptical distributions

Most of the current approaches to mixture modeling consider mixture components from a few families of probability distributions, in particular from the Gaussian family. The reasons of these preferences can be traced to their training algorithms, typically versions of the Expectation-Maximization (EM) method. The re-estimation equations needed by this method become very complex as the mixture components depart from the simplest cases. Here we propose to use a stochastic approximation method for probabilistic mixture learning. Under this method it is straightforward to train mixtures composed by a wide range of mixture components from different families. Hence, it is a flexible alternative for mixture learning. Experimental results are presented to show the probability density and missing value estimation capabilities of our proposal.     

Set-membership estimation from poor quality data sets: Modelling ammonia volatilisation in flooded rice systems

A set-membership (bounded-error) estimation approach can handle small and poor quality data sets as it does not require testing of statistical assumptions which is possible only with large informative data sets. Thus, set-membership estimation can be a good tool in the modelling of agri-environmental systems, which typically suffers from limited and poor quality observational data sets. The objectives of the paper are (i) to demonstrate how six parameters in an agri-environmental model, developed to estimate NH₃ volatilisation in flooded rice systems, were estimated based on two data sets using a set-membership approach, and (ii) to compare the set-membership approach with conventional non-linear least-squares methods. Results showed that the set-membership approach is efficient in retrieving feasible parameter-vectors compared with non-linear least-squares methods. The set of feasible parameter-vectors allows the formation of a dispersion matrix of which the eigenvalue decomposition reflects the parameter sensitivity in a region.     

Frequency domain regularization of d-dimensional structure tensor-based directional fields

The present work is intended to address two of the major difficulties that can be found when tackling the estimation of the local orientation of the data in a scene, a task which is usually accomplished by means of the computation of the structure tensor-based directional field. On one hand, the orientation information only exists in the non-homogeneous regions of the dataset, while it is zero in the areas where the gradient (i.e. the first-order intensity variation) remains constant. Due to this lack of information, there are many cases in which the overall shape of the represented objects cannot be precisely inferred from the directional field. On the other hand, the orientation estimation is highly dependent on the particular choice of the averaging window used for its computation (since a collection of neighboring gradient vectors is needed to obtain a dominant orientation), typically resulting in vector fields which vary from very irregular (thus yielding a noisy estimation) to very uniform (but at the expense of a loss of angular resolution). The proposed solution to both drawbacks is the regularization of the directional field; this process extends smoothly the previously computed vectors to the whole dataset while preserving the angular information of relevant structures. With this purpose, the paper introduces a suitable mathematical framework and deals with the d-dimensional variational formulation which is derived from it. The proposed formulation is finally translated into the frequency domain in order to obtain an increase of insight on the regularization problem, which can be understood as a low-pass filtering of the directional field. The frequency domain point of view also allows for an efficient implementation of the resulting iterative algorithm. Simulation experiments involving datasets of different dimensionality prove the validity of the theoretical approach.     

Guaranteed Dynamic-System State Estimation by the Method of Bidirectional Dynamic Programming

The problem of guaranteed estimation (smoothing, filtration, prediction) of a dynamic process observed on a finite discrete time interval is solved, based on generalization of the dynamic programming procedure for the case with sequential optmization in direct and inverse time.     

Estimation of the Bivariate Stable Spectral Representation by theProjection Method

A method of estimating the spectral representation of a generalized bivariatestable distribution is presented, based on a series of maximum likelihood (ML)estimates of the stable parameters of univariate projections of the data. Thecorresponding stable spectral density is obtained by solving a quadraticprogram. The proposed method avoids the often arduous task of computing themultivariate stable density, relying instead on the standard univariate stabledensity. The paper applies this projection procedure, under the simplifyingassumption of symmetry, to simulated data as well as to foreign exchangereturn data, with favorable results. Kanter projection coefficients governingconditional expectations are computed from the estimated spectral density. For the simulated data these compare well to their known true values.     

An Artificial Neural Network Classifier Design Based-on Variable Kernel and Non-Parametric Density Estimation

This paper proposes a probabilistic variant of the SOM-kMER (Self Organising Map-kernel-based Maximum Entropy learning Rule) model for data classification. The classifier, known as pSOM-kMER (probabilistic SOM-kMER), is able to operate in a probabilistic environment and to implement the principles of statistical decision theory in undertaking classification problems. A distinctive feature of pSOM-kMER is its ability in revealing the underlying structure of data. In addition, the Receptive Field (RF) regions generated can be used for variable kernel and non-parametric density estimation. Empirical evaluation using benchmark datasets shows that pSOM-kMER is able to achieve good performance as compared with those from a number of machine learning systems. The applicability of the proposed model as a useful data classifier is also demonstrated with a real-world medical data classification problem.     

Some Statistical-Estimation Methods for Stochastic Finite-State Transducers

Formal translations constitute a suitable framework for dealing with many problems in pattern recognition and computational linguistics. The application of formal transducers to these areas requires a stochastic extension for dealing with noisy, distorted patterns with high variability. In this paper, some estimation criteria are proposed and developed for the parameter estimation of regular syntax-directed translation schemata. These criteria are: maximum likelihood estimation, minimum conditional entropy estimation and conditional maximum likelihood estimation. The last two criteria were proposed in order to deal with situations when training data is sparse. These criteria take into account the possibility of ambiguity in the translations: i.e., there can be different output strings for a single input string. In this case, the final goal of the stochastic framework is to find the highest probability translation of a given input string. These criteria were tested on a translation task which has a high degree of ambiguity.     

ON THE PARTIAL SUMS OF RESIDUALS IN AUTOREGRESSIVE AND MOVING AVERAGE MODELS

Abstract. The limiting process of partial sums of residuals in stationary and invertible autoregressive moving-average models is studied. It is shown that the partial sums converge to a standard Brownian motion under the assumptions that estimators of unknown parameters are root- n consistent and that innovations are independent and identically distributed random variables with zero mean and finite variance or, more generally, are martingale differences with moment restrictions specified in Theorem 1. Applications for goodness-of-fit and change-point problems are considered. The use of residuals for constructing nonparametric density estimation is discussed.     

Accurate mirror-based camera pose estimation with explicit geometric meanings

An accurate closed form solution is proposed to estimate camera pose by several mirrored reference object images acquired via a planar mirror under different unknown poses. Compared with state-of-the-art methods, our method is more accurate when there are more than three images and has explicit geometric meanings. This method also properly handles cases in which some of the mirror poses are parallel. The central idea is to minimize an error metric based on all reflections of rotation, which enables the camera rotation to be estimated directly by SVD of sum of mirrored camera rotations. After that, the camera translation is computed by solving a large system of linear equations to minimize object space collinearity error. Both synthesized data and real data experiments show the advantages of our approach.