The paper is concerned with the asymptotic normality of a certain statistic based on the logarithms of disjoint m-spacings. The exact and asymptotic mean and variance are computed in the case of uniform distribution on the interval [0,1]. This result is generalized to the case when the sample is drawn from a distribution with positive step density on [0,1].
We introduce a quantitative measure Δ of stability in optimal sequential testing of two simple hypotheses about a density of observations: f=f₀ versus f=f₁. The index Δ represents an additional cost paid when a stopping rule optimal for the pair (f₀,f₁) is applied to test the hypothesis f=f₀ versus a "perturbed alternative" f=f̃₁. An upper bound for Δ is established in terms of the total variation distance between f₁(X)/f₀(X) and f̃₁(X)/f₀(X) with X∼f₀.
We study the universal estimator for the regression problem in learning theory considered by Binev et al. This new approach allows us to improve their results.
The paper gives an overview of feature selection techniques in statistical pattern recognition with particular emphasis on methods developed within the Institute of Information Theory and Automation research team throughout recent years. Besides discussing the advances in methodology since times of Perez’s pioneering work the paper attempts to put the methods into a taxonomical framework. The methods discussed include the latest variants of the optimal algorithms, enhanced sub-optimal techniques...
En este trabajo se presenta una metodología que permite clasificar funciones de distribución absolutamente continuas unidimensionales atendiendo a sus ramas. La idea básica es que, en las ramas la función de distribución difiere en un infinitésimo del valor uno o cero dependiendo de la rama de interés. La principal ventaja de esta clasificación es su aplicación a la teoría de distribuciones de extremos. En esta línea se obtienen nuevas familias de distribuciones de extremos. Entre ellas, las clásicas...
This work deals with a multivariate random coefficient autoregressive model (RCA) of the first order. A class of modified least-squares estimators of the parameters of the model, originally proposed by Schick for univariate first-order RCA models, is studied under more general conditions. Asymptotic behavior of such estimators is explored, and a lower bound for the asymptotic variance matrix of the estimator of the mean of random coefficient is established. Finite sample properties are demonstrated...
Bayesian estimation for the two parameters of a Gumbel distribution are obtained based on kth lower record values. Prediction, either point or interval, for future kth lower record values is also presented from a Bayesian view point. Some of the results of [4] can be obtained as special cases of our results (k=1).
A robust significance testing method for the Cox regression model, based on a modified Wald test statistic, is discussed. Using Monte Carlo experiments the asymptotic behavior of the modified robust versions of the Wald statistic is compared with the standard significance test for the Cox model based on the log likelihood ratio test statistic.
A kernel estimator of the squared -norm of the intensity function of a Poisson random field is defined. It is proved that the estimator is asymptotically unbiased and strongly consistent. The problem of estimating the squared -norm of a function disturbed by a Wiener random field is also considered.
In the problem of signal detection in gaussian white noise we show asymptotic minimaxity of kernel-based tests. The test statistics equal -norms of kernel estimates. The sets of alternatives are essentially nonparametric and are defined as the sets of all signals such that the -norms of signal smoothed by the kernels exceed some constants . The constant depends on the power of noise and as . Similar statements are proved also if an additional information on a signal smoothness is given....
In the problem of signal detection in Gaussian white noise we show asymptotic minimaxity of kernel-based tests. The test statistics equal L2-norms of kernel estimates. The sets of alternatives are essentially nonparametric and are defined as the sets of all signals such that the L2-norms of signal smoothed by the kernels exceed some constants pε > 0. The constant pε depends on the power ϵ of noise and pε → 0 as ε → 0. Similar statements are proved also if an additional information on a signal...
We study the impact of certain transformations within the class of Archimedean copulas. We give some admissibility conditions for these transformations, and define some equivalence classes for both transformations and generators of Archimedean copulas. We extend the r-fold composition of the diagonal section of a copula, from r ∈ N to r ∈ R. This extension, coupled with results on equivalence classes, gives us new expressions of transformations and generators. Estimators deriving directly from these...