Nonparametric time series regression

Consider ak-times differentiable unknown regression function(·) of ad-dimensional measurement variable. LetT() denote a derivative of(·) of orderm and setr=(k–m)/(2k+d). Given a bivariate stationary time series of lengthn, under some appropriate conditions, a sequence of local polynomial estimators of the functionT() can be chosen to achieve the optimal rate of convergencen ^–r inL ₂ norms restricted to compacts; and the optimal rate (n ^–1 logn) ^r in theL norms on compacts. These results generalize those by Stone (1982,Ann. Statist.,10, 1040–1053) which deals with nonparametric regression estimation for random (i.i.d.) samples. Applications of these results to nonlinear time series problems will also be discussed.This work was completed while the author was visiting Mathematical Sciences Research Institute at Berkeley, California. Research was supported in part by NSF Grant DMS-8505550, NC Board of Science and Technology Development Award 90SE06 and UNC Research Council.     

漂移向量与扩散矩阵的非参数估计模型

考虑多维扩散过程的非参数估计问题.利用It扩散的性质,将漂移向量和扩散矩阵的样本表示成带有测量误差的回归模型,并讨论了系统误差的L~r上界以及随机误差项的收敛速度,建立了漂移向量与扩散矩阵非参数估计的通用模型.     

On periodically isotonic climate change

In a view of the past two millennia, is the recent 160-year global warmth an anomaly or a continuation of past climate pattern? In this work, we present two non-parametric χ² tests based on the U-test for the isotonic change-point problem proposed in an unpublished work by G. Shen and H. Xu (henceforth abbreviated as SX (unpublished)) and apply them to the global warming data. Our tests do not require any structure assumption on the underlying the deterministic mean process {μ_t} under the periodically isotonic alternative, but accommodate a large class of the models for the random error process. Being more attractive, the two tests obviate estimation of {μ_t} and possess explicit asymptotic distributions for short-range dependence data, and have asymptotic power 1 even under the local alternative.     

Nonasymptotic Bounds on the L 2 Error of Neural Network Regression Estimates

The estimation of multivariate regression functions from bounded i.i.d. data is considered. The L ₂ error with integration with respect to the design measure is used as an error criterion. The distribution of the design is assumed to be concentrated on a finite set. Neural network estimates are defined by minimizing the empirical L ₂ risk over various sets of feedforward neural networks. Nonasymptotic bounds on the L ₂ error of these estimates are presented. The results imply that neural networks are able to adapt to additive regression functions and to regression functions which are a sum of ridge functions, and hence are able to circumvent the curse of dimensionality in these cases.     

Empirical Likelihood Confidence Bands in Density Estimation

Abstract

Empirical likelihood methods are developed for constructing confidence bands in problems of nonparametric density estimation. These techniques have an advantage over more conventional methods in that the shape of the bands is determined solely by the data. We show how to construct an empirical likelihood functional, rather than a function, and contour it to produce the confidence bands. Analogs of Wilks's theorem are established in this infinite-parameter setting and may be used to select the appropriate contour. An alternative calibration, based on the bootstrap, is also suggested. Large-sample theory is developed to show that the bands have asymptotically correct coverage, and a numerical example is presented to demonstrate the technique. Comparisons are made with the use of bootstrap replications to choose both the shape and size of the bands.     

Wald意义下的局部π-Bayes最优决策

本文在某些必要的适当条件下,采用一种独特的方式证明Wald意义下的最优决策的存在性.并表明:Wald意义下的最优决策作为一种π-Bayes最优决策实际上是局部的.     

Improving Penalized Least Squares through Adaptive Selection of Penalty and Shrinkage

Estimation of the mean function in nonparametric regression is usefully separated into estimating the means at the observed factor levels—a one-way layout problem—and interpolation between the estimated means at adjacent factor levels. Candidate penalized least squares (PLS) estimators for the mean vector of a one-way layout are expressed as shrinkage estimators relative to an orthogonal regression basis determined by the penalty matrix. The shrinkage representation of PLS suggests a larger class of candidate monotone shrinkage (MS) estimators. Adaptive PLS and MS estimators choose the shrinkage vector and penalty matrix to minimize estimated risk. The actual risks of shrinkage-adaptive estimators depend strongly upon the economy of the penalty basis in representing the unknown mean vector. Local annihilators of polynomials, among them difference operators, generate penalty bases that are economical in a range of examples. Diagnostic techniques for adaptive PLS or MS estimators include basis-economy plots and estimates of loss or risk.     

Nonparametric Multivariate CUSUM Chart and EWMA Chart Based on Simplicial Data Depth

Simplicial data depth is a useful tool for describing how central a vector is in a multivariate distribution. If the average simplicial depth of a subgroup of observations from a multivariate distribution is too small, it may indicate that a shift in its location or/both scale occurs. In this paper, we propose two new types of nonparametric control charts which are one-sided CUSUM and EWMA control schemes based on simplicial data depth. We also compute the Average Run Length of the CUSUM chart and the EWMA chart by Markov chain method. Recommendations on how to choose the optimal reference value and the smoothing parameter are also given. Comparisons between these two proposed control schemes and the multivariate EWMA are presented.     

Spline Multiscale Smoothing to Control FDR for Exploring Features of Regression Curves

SiZer (significant zero crossing of the derivatives) is a multiscale smoothing method for exploring trends, maxima, and minima in data. In this article, a regression spline version of SiZer is proposed in a nonparametric regression setting by the fiducial method. The number of knots for spline interpolation is used as the scale parameter of the new SiZer, which controls the smoothness of estimate. In the construction of the new SiZer, multiple testing adjustment is made to control the row-wise false discovery rate (FDR) of SiZer. This adjustment is appealing for exploratory data analysis and has potential to increase the power. A special map is also produced on a continuous scale using p-values to assess the significance of features. Simulations and a real data application are carried out to investigate the performance of the proposed SiZer, in which several comparisons with other existing SiZers are presented. Supplementary materials for this article are available online.     

A nonparametric variable step-size NLMS algorithm for transversal filters

A nonparametric adaptive filtering approach is proposed in this paper. The algorithm is obtained by exploiting a time-varying step size in the traditional NLMS weight update equation. The step size is adjusted according to the square of a time-averaging estimate of the autocorrelation of a priori and a posteriori error. Therefore, the new algorithm has more effective sense proximity to the optimum solution independent of uncorrelated measurement noise. Moreover, this algorithm has fast convergence at the early stages of adaptation and small final misadjustment at steady-state process. It works reliably and is easy to implement since the update function is nonparametric. Furthermore, the experimental results in system identification applications are presented to illustrate the principle and efficiency of the proposed algorithm.