Some notes on extremal discriminant analysis

Classical discriminant analysis focusses on Gaussian and nonparametric models where in the second case the unknown densities are replaced by kernel densities based on the training sample. In the present article we assume that it suffices to base the classification on exceedances above higher thresholds, which can be interpreted as observations in a conditional framework. Therefore, the statistical modeling of truncated distributions is merely required. In this context, a nonparametric modeling is not adequate because the kernel method is inaccurate in the upper tail region. Yet one may deal with truncated parametric distributions like the Gaussian ones. Our primary aim is to replace truncated Gaussian distributions by appropriate generalized Pareto distributions and to explore properties and the relationship of discriminant functions in both models.     

Bayesian Semiparametric Density Deconvolution in the Presence of Conditionally Heteroscedastic Measurement Errors

We consider the problem of estimating the density of a random variable when precise measurements on the variable are not available, but replicated proxies contaminated with measurement error are available for sufficiently many subjects. Under the assumption of additive measurement errors this reduces to a problem of deconvolution of densities. Deconvolution methods often make restrictive and unrealistic assumptions about the density of interest and the distribution of measurement errors, for example, normality and homoscedasticity and thus independence from the variable of interest. This article relaxes these assumptions and introduces novel Bayesian semiparametric methodology based on Dirichlet process mixture models for robust deconvolution of densities in the presence of conditionally heteroscedastic measurement errors. In particular, the models can adapt to asymmetry, heavy tails, and multimodality. In simulation experiments, we show that our methods vastly outperform a recent Bayesian approach based on estimating the densities via mixtures of splines. We apply our methods to data from nutritional epidemiology. Even in the special case when the measurement errors are homoscedastic, our methodology is novel and dominates other methods that have been proposed previously. Additional simulation results, instructions on getting access to the dataset and R programs implementing our methods are included as part of online supplementary materials.     

Likelihood Inference under Generalized Hybrid Censoring Scheme with Comp eting Risks

Statistical inference is developed for the analysis of generalized type-II hybrid censoring data under exponential competing risks model. In order to solve the problem that approximate methods make unsatisfactory performances in the case of small sample size, we establish the exact conditional distributions of estimators for parameters by conditional moment generating function(CMGF). Furthermore, confidence intervals(CIs) are constructed by exact distributions, approximate distributions as well as bootstrap method respectively, and their performances are evaluated by Monte Carlo simulations. And finally, a real data set is analyzed to illustrate all the methods developed here.     

The Generalized Lognormal Distribution and the Stieltjes Moment Problem

This paper studies a Stieltjes-type moment problem defined by the generalized lognormal distribution, a heavy-tailed distribution with applications in economics, finance, and related fields. It arises as the distribution of the exponential of a random variable following a generalized error distribution, and hence figures prominently in the exponential general autoregressive conditional heteroskedastic (EGARCH) model of asset price volatility. Compared to the classical lognormal distribution it has an additional shape parameter. It emerges that moment (in)determinacy depends on the value of this parameter: for some values, the distribution does not have finite moments of all orders, hence the moment problem is not of interest in these cases. For other values, the distribution has moments of all orders, yet it is moment-indeterminate. Finally, a limiting case is supported on a bounded interval, and hence determined by its moments. For those generalized lognormal distributions that are moment-indeterminate, Stieltjes classes of moment-equivalent distributions are presented.     

Compatibility of conditionally specified models

A conditionally specified joint model is convenient to use in fields such as spatial data modeling, Gibbs sampling, and missing data imputation. One potential problem with such an approach is that the conditionally specified models may be incompatible, which can lead to serious problems in applications. We propose an odds ratio representation of a joint density to study the issue and derive conditions under which conditionally specified distributions are compatible and yield a joint distribution. Our conditions are the simplest to verify compared with those proposed in the literature. The proposal also explicitly constructs joint densities that are fully compatible with the conditionally specified densities when the conditional densities are compatible, and partially compatible with the conditional densities when they are incompatible. The construction result is then applied to checking the compatibility of the conditionally specified models. Ways to modify the conditionally specified models based on the construction of the joint models are also discussed when the conditionally specified models are incompatible.     

Regularized fast multiple-image deconvolution for LBT

In this paper we study a fast deconvolution technique for the image restoration problem of the Large Binocular Telescope (LBT) interferometer. Since LBT provides several blurred and noisy images of the same object, it requires the use of multiple-image deconvolution methods in order to produce a unique image with high resolution. Hence the restoration process is basically a linear ill-posed problem, with overdetermined system and data corrupted by several components of noise.     

Some bivariate gamma distributions

We introduce two new bivariate gamma distributions based on a characterizing property involving products of gamma and beta random variables. We derive various representations for their joint densities, product moments, conditional densities and conditional moments. Some of these representations involve special functions such as the complementary incomplete gamma and Whittaker functions. We also discuss ways to construct multivariate generalizations.     

On optimal parameter estimation in credibility

The present paper is concerned with optimal estimation of the 1st and 2nd order structural moments appearing in credibility formulas. In a recent paper De Vylder has treated the problem in the case of multinormal conditional distributions under quite restrictive assumptions. He minimizes, within a certain restricted class of unbiased estimators, the variance (or the sum of variances if the estimand is a matrix) and next replaces all structural moments (up to fourth order) in the solution by estimates based on the data. This paper is an attempt to simplify the method and extend it so as to make it applicable in more general situations. By suitable choice of a (sufficient) set of statistics and a suitable parametrization, the powerful theory of estimation in linear models can be employed, which makes cumbersome minimization procedures superfluous. The theory is applied to the cases with binomial. Poisson, compound Poisson, and multinormal conditional distributions. Some simulation studies have been performed to assess the performance of the estimators.     

Deconvolving compactly supported densities

This paper addresses the statistical problem of density deconvolution under the condition that the density to be estimated has compact support. We introduce a new estimation procedure, which establishes faster rates of convergence for smooth densities as compared to the optimal rates for smooth densities with unbounded support. This framework also allows us to relax the usual condition of known error density with non-vanishing Fourier transform, so that a nonparametric class of densities is valid; therefore, even the shape of the noise density need not be assumed. These results can also be generalized for fast decaying densities with unbounded support. We prove optimality of the rates in the underlying experiment and study the practical performance of our estimator by numerical simulations.      

Generalized symmetrized dirichlet distributions

In this paper we introduce a family of multivariate distributions, which consists of scale mixtures of symmetrized Dirichlet distributions. This family is a symmetrization of multivariate Liouville distributions and contains the well-known spherically symmetric distributions as a special case. The basic properties of this family such as stochastic representation, probability density functions, marginal and conditional distributions and components' independence are studied. A criterion of the invariance of statistics is also given.     

Sur la convergence de l'estimation conditionnelle itérative

The iterative conditional estimation (ICE) is an iterative estimation method of the parameters in the case of incomplete data. Its use asks for relatively weak hypotheses and it can be performed in relatively complex situations, as in triplet Markov models. The aim of this Note is to express a general theorem of convergence of ICE, and to show its applicability in the problem of the estimation of the proportions in a mixture of multivariate distributions. To cite this article: W. Pieczynski, C. R. Acad. Sci. Paris, Ser. I 346 (2008).     

The Construction of Multivariate Distributions from Markov Random Fields

We address the problem of constructing and identifying a valid joint probability density function from a set of specified conditional densities. The approach taken is based on the development of relations between the joint and the conditional densities using Markov random fields (MRFs). We give a necessary and sufficient condition on the support sets of the random variables to allow these relations to be developed. This condition, which we call the Markov random field support condition, supercedes a common assumption known generally as the positivity condition. We show how these relations may be used in reverse order to construct a valid model from specification of conditional densities alone. The constructive process and the role of conditions needed for its application are illustrated with several examples, including MRFs with multiway dependence and a spatial beta process.     

Multiway Dependence in Exponential Family Conditional Distributions

Conditionally specified statistical models are frequently constructed from one-parameter exponential family conditional distributions. One way to formulate such a model is to specify the dependence structure among random variables through the use of a Markov random field (MRF). A common assumption on the Gibbsian form of the MRF model is that dependence is expressed only through pairs of random variables, which we refer to as the “pairwise-only dependence” assumption. Based on this assumption, J. Besag (1974, J. Roy. Statist. Soc. Ser. B36, 192–225) formulated exponential family “auto-models” and showed the form that one-parameter exponential family conditional densities must take in such models. We extend these results by relaxing the pairwise-only dependence assumption, and we give a necessary form that one-parameter exponential family conditional densities must take under more general conditions of multiway dependence. Data on the spatial distribution of the European corn borer larvae are fitted using a model with Bernoulli conditional distributions and several dependence structures, including pairwise-only, three-way, and four-way dependencies.     

Convergence guarantees for generalized adaptive stochastic search methods for continuous global optimization

This paper presents some simple technical conditions that guarantee the convergence of a general class of adaptive stochastic global optimization algorithms. By imposing some conditions on the probability distributions that generate the iterates, these stochastic algorithms can be shown to converge to the global optimum in a probabilistic sense. These results also apply to global optimization algorithms that combine local and global stochastic search strategies and also those algorithms that combine deterministic and stochastic search strategies. This makes the results applicable to a wide range of global optimization algorithms that are useful in practice. Moreover, this paper provides convergence conditions involving the conditional densities of the random vector iterates that are easy to verify in practice. It also provides some convergence conditions in the special case when the iterates are generated by elliptical distributions such as the multivariate Normal and Cauchy distributions. These results are then used to prove the convergence of some practical stochastic global optimization algorithms, including an evolutionary programming algorithm. In addition, this paper introduces the notion of a stochastic algorithm being probabilistically dense in the domain of the function and shows that, under simple assumptions, this is equivalent to seeing any point in the domain with probability 1. This, in turn, is equivalent to almost sure convergence to the global minimum. Finally, some simple results on convergence rates are also proved.     

A Multisensor Deconvolution Problem

Meisters and Peterson gave an equivalent condition under which the multisensor deconvolution problem has a solution when there are two convolvers, each the characteristic function of an interval. In this article we find additional conditions under which the deconvolution problem for multiple characteristic functions is solvable. We extend the result to the space of Gevrey distributions and prove that every linear operator S, fromthe space of Gevrey functions with compact support onto itself, which commutes with translations can be represented as convolution with a unique Gevrey distribution T of compact support. Finally, we find explicit formula for deconvolvers when the convolvers satisfy weaker conditions than the equivalence conditions using nonperiodic sampling method.     

Numerical method for estimating multivariate conditional distributions

Summary  A computational framework for estimation of multivariate conditional distributions is presented. It allows the forecast of the joint distribution of target variables in dependence on explaining variables. The concept can be applied to general distribution families such as stable or hyperbolic distributions. The estimation is based on the numerical minimization of the cross entropy, using the Multi-Level Single-Linkage global optimization method. Nonlinear dependencies of conditional parameters can be modeled with help of general functional approximators such as multi-layer perceptrons. In applications, the information about a complete distribution of forecasts can be used to quantify the reliability of the forecast or for decision support. This is illustrated on a case study concerning the spare parts demand forecast. The improvement of the forecast error due to using non-Gaussian distributions is presented in another case study concerning the truck sales forecast.     

Prediction and Inference for Truncated Spatial Data

Abstract

Spatial data in mining, hydrology, and pollution monitoring commonly have a substantial proportion of zeros. One way to model such data is to suppose that some pointwise transformation of the observations follows the law of a truncated Gaussian random field. This article considers Monte Carlo methods for prediction and inference problems based on this model. In particular, a method for computing the conditional distribution of the random field at an unobserved location, given the data, is described. These results are compared to those obtained by simple kriging and indicator cokriging. Simple kriging is shown to give highly misleading results about conditional distributions; indicator cokriging does quite a bit better but still can give answers that are substantially different from the conditional distributions. A slight modification of this basic technique is developed for calculating the likelihood function for such models, which provides a method for computing maximum likelihood estimates of unknown parameters and Bayesian predictive distributions for values of the process at unobserved locations.     

Equivalent Conditions of Asymptotics for the Density of the Supremum of a Random Walk in the Intermediate Case

This paper obtains some equivalent conditions about the asymptotics for the density of the supremum of a random walk with light-tailed increments in the intermediate case. To do this, the paper first corrects the proofs of some existing results about densities of random sums. On the basis of the above results, the paper obtains some equivalent conditions about the asymptotics for densities of ruin distributions in the intermediate case and densities of infinitely divisible distributions. In the above studies, some differences and relations between the results on a distribution and its corresponding density can be discovered.      

Bivariate Distributions with Pearson Type VII Conditionals

In this article the most general class of bivariate distributions such that both conditional densities are Pearson Type VII, with fixed shape parameter, is fully characterized. Some of its properties and relations with other distributions are explored. The estimation of parameters is considered by the methods of maximum likelihood and pseudolikelihood and a method for random variate generation is presented along with a simulation experiment. Bivariate and multivariate extensions of the Pearson Type VII conditionals distribution are also discussed.     

Transition Law-based Simulation of Generalized Inverse Gaussian Ornstein–Uhlenbeck Processes

In this paper, a stochastic integral of Ornstein–Uhlenbeck type is represented to be the sum of three independent random variables—one follows a distribution whose density is a deconvolution of the densities of two generalized inverse Gaussian distributions, and the two others all have compound Poisson distributions. Based on the representation of the stochastic integral, a simulation procedure for obtaining discretely observed values of Ornstein–Uhlenbeck processes with given generalized inverse Gaussian distribution is provided. For some subclasses of the generalized inverse Gaussian Ornstein–Uhlenbeck process, the innovations can be sampled exactly. The performance of the simulation method is evidenced by some empirical results.