The adaptive LASSO spline estimation of single-index model

In this paper, based on spline approximation, the authors propose a unified variable selection approach for single-index model via adaptive L ₁ penalty. The calculation methods of the proposed estimators are given on the basis of the known lars algorithm. Under some regular conditions, the authors demonstrate the asymptotic properties of the proposed estimators and the oracle properties of adaptive LASSO (aLASSO) variable selection. Simulations are used to investigate the performances of the proposed estimator and illustrate that it is effective for simultaneous variable selection as well as estimation of the single-index models.     

An M-Objective Penalty Function Algorithm Under Big Penalty Parameters

Some classical penalty function algorithms may not always be convergent under big penalty parameters in Matlab software, which makes them impossible to find out an optimal solution to constrained optimization problems. In this paper, a novel penalty function (called M-objective penalty function) with one penalty parameter added to both objective and constrained functions of inequality constrained optimization problems is proposed. Based on the M-objective penalty function, an algorithm is developed to solve an optimal solution to the inequality constrained optimization problems, with its convergence proved under some conditions. Furthermore, numerical results show that the proposed algorithm has a much better convergence than the classical penalty function algorithms under big penalty parameters, and is efficient in choosing a penalty parameter in a large range in Matlab software.     

Variable selection in joint location,scale and skewness models of the skew-normal distribution

Variable selection is an important research topic in modern statistics, traditional variable selection methods can only select the mean model and (or) the variance model, and cannot be used to select the joint mean, variance and skewness models. In this paper, the authors propose the joint location, scale and skewness models when the data set under consideration involves asymmetric outcomes, and consider the problem of variable selection for our proposed models. Based on an efficient unified penalized likelihood method, the consistency and the oracle property of the penalized estimators are established. The authors develop the variable selection procedure for the proposed joint models, which can efficiently simultaneously estimate and select important variables in location model, scale model and skewness model. Simulation studies and body mass index data analysis are presented to illustrate the proposed methods.     

Single-Index Quantile Regression with Left Truncated Data

The purpose of this paper is two fold. First, the authors investigate quantile regression (QR) estimation for single-index QR models when the response is subject to random left truncation. The random weights are introduced to deal with left truncated data and the associated iteration estimation method is proposed. The asymptotic properties for the proposed QR estimates of the index parameter and unknown link function are both obtained. Further, by combining the QR loss function and the adaptive LASSO penalization, a variable selection procedure for the index parameter is introduced and its oracle property is established. Second, a weighted empirical log-likelihood ratio of the index parameter based on the QR method is introduced and is proved to be asymptotic standard chi-square distribution. Furthermore, confidence regions of the index parameter can be constructed. The finite sample performance of the proposed methods are demonstrated. A real data analysis is also conducted to show the usefulness of the proposed approaches.

     

MODEL SELECTION FOR LOG-LINEAR MODELS OF CONTINGENCY TABLES

In this paper, we propose an information-theoretic-criterion-based model selection procedure for log-linear model of contingency tables under multinomial sampling, and establish the strong consistency of the method under some mild conditions. An exponential bound of miss detection probability is also obtained. The .selection procedure is modified so that it can be used in practice. Simulation shows that the modified method is valid. To avoid selecting the penalty coefficient in the information criteria, an alteruative selection procedure is given.     

Sequential Estimate for Generalized Linear Models with Uncertain Number of Effective Variables

For the generalized linear model,the authors propose a sequential sampling procedure based on an adaptive shrinkage estimate of parameter.This method can determine a minimum sample size under which effective variables contributing to the model are identified and estimates of regression parameters achieve the required accuracy.The authors prove that the proposed sequential procedure is asymptotically optimal.Numerical simulation studies show that the proposed method can save a large number of samples compared to the traditional sequential approach.     

Variable Selection for Structural Equation with Endogeneity

This paper studies variable selection problem in structural equation of a two-stage least squares (2SLS) model in presence of endogeneity which is commonly encountered in empirical economic studies. Model uncertainty and variable selection in the structural equation is an important issue as described in Andrews and Lu (2001) and Caner (2009). The authors propose an adaptive Lasso 2SLS estimator for linear structural equation with endogeneity and show that it enjoys the oracle properties, i.e., the consistency in both estimation and model selection. In Monte Carlo simulations, the authors demonstrate that the proposed estimator has smaller bias and MSE compared with the bridge-type GMM estimator (Caner, 2009). In a case study, the authors revisit the classic returns to education problem (Angrist and Krueger, 1991) using the China Population census data. The authors find that the education level not only has strong effects on income but also shows heterogeneity in different age cohorts.     

Distributed Algorithm for Robust Resource Allocation with Polyhedral Uncertain Allocation Parameters

This paper studies a distributed robust resource allocation problem with nonsmooth objective functions under polyhedral uncertain allocation parameters. In the considered distributed robust resource allocation problem, the (nonsmooth) objective function is a sum of local convex objective functions assigned to agents in a multi-agent network. Each agent has a private feasible set and decides a local variable, and all the local variables are coupled with a global affine inequality constraint, which is subject to polyhedral uncertain parameters. With the duality theory of convex optimization, the authors derive a robust counterpart of the robust resource allocation problem. Based on the robust counterpart, the authors propose a novel distributed continuous-time algorithm, in which each agent only knows its local objective function, local uncertainty parameter, local constraint set, and its neighbors’ information. Using the stability theory of differential inclusions, the authors show that the algorithm is able to find the optimal solution under some mild conditions. Finally, the authors give an example to illustrate the efficacy of the proposed algorithm.     

变宽度凸组合变阶数LMS自适应滤波算法

为了进一步改善凸组合变阶数最小均方(convex combination variable fractional tap-length least mean square,CFTLMS)自适应滤波算法的稳态性能,在证明其稳态性能的基础上,提出了一种变宽度凸组合变阶数(variable width-CFTLMS, VW-CFTLMS)自适应滤波算法,并给出参数的选择依据。仿真结果验证了低信噪比情况下,VW-CFTLMS算法稳态性能和参数选择依据的正确性;同时该算法的稳态性能要优于CFTLMS算法,其额外均方误差相比于CFTLMS算法降低约1.8 dB,具有实用价值。     

Unbounded batch scheduling with a common due window on a single machine*

The common due window scheduling problem with batching on a single machine is dealt with to minimize the total penalty of weighted earliness and tardiness. In this paper it is assumed that a job incurs no penalty as long as it is completed within the common due window. It is the first time for the due window scheduling to be extended to this situation so that jobs can be processed in batches. An unbounded version of batch scheduling is also considered. Hence, jobs, no matter how many there are, can be processed in a batch once the machine is free. For two cases that the location of due window is either a decision variable or a given parameter, polynomial algorithms are proposed based on several optimal properties.     

基于自适应变异PSO-SVM的APU性能参数预测模型

为提高辅助动力装置(auxiliary power unit, APU)性能参数预测的精度,针对支持向量机(support vector machine, SVM)模型在实际使用中遇到的参数选择问题,采用自适应变异粒子群优化(particle swarm optimization, PSO)算法实现对SVM惩罚参数和核参数的优化选择,提出一种基于自适应变异PSO算法优化SVM的APU性能参数预测模型。进一步分析了预测模型不同预测步长对短期预测精度的影响。利用某型APU性能参数数据进行了验证,并与多种预测模型进行了对比实验。实验结果表明,对于排气温度(exhaust gas temperature, EGT)的预测,自适应变异PSO-SVM模型的平均绝对百分比误差(mean absolute percentage error, MAPE)比标准PSO-SVM模型低47%;对于滑油温度(oil temperrature, OT)的预测,自适应变异PSO-SVM模型的MAPE比标准PSO-SVM低29%,为短期APU性能变化趋势预测提供了一定的参考。     

Feature Screening for Nonparametric and Semiparametric Models with Ultrahigh-Dimensional Covariates

This paper considers the feature screening and variable selection for ultrahigh dimensional covariates. The new feature screening procedure base on conditional expectation which is used to differentiate whether an explanatory variable contributes to a response variable or not, without requiring a specific parametric form of the underlying data model. The authors estimate the marginal conditional expectation by kernel regression estimator. The proposed method is showed to have sure screen property. The authors propose an iterative kernel estimator algorithm to reduce the ultrahigh dimensionality to an appropriate scale. Simulation results and real data analysis demonstrate the proposed method works well and performs better than competing methods.     

Statistical Diagnosis for HIV Dynamics Based on Mean Shift Outlier Model

Ordinary differential equation(ODE) are widely used for quantifying HIV viral dynamics.It is interesting but challenging to estimate ODE parameters from noisy data, especially when the data have some outliers. In this study, the authors use the Mean Shift Outlier Model(MSOM) to detect outliers in HIV model based on the two-step estimation of ODE. Approximate formula for shift parameter is derived. Furthermore, a score test statistic is constructed and its approximating distribution is established. The simulation results show that: 1) The boundary points have more impact on the parameter estimation relative to interior points. 2) The proposed procedure can detect the outliers effectively. The authors illustrate the proposed approach using an application example from an HIV clinical trial and find similar pattern to the simulation studies.     

Four step scheme for general Markovian forward-backward SDES

<正> This paper studies a class of forward-backward stochastic differential equations (FBSDE)in a general Markovian framework.The forward SDE represents a large class of strong Markov semimartingales,and the backward generator requires only mild regularity assumptions.The authors showthat the Four Step Scheme introduced by Ma,et al.(1994) is still effective in this case.Namely,the authors show that the adapted solution of the FBSDE exists and is unique over any prescribedtime duration;and the backward components can be determined explicitly by the forward componentvia the classical solution to a system of parabolic integro-partial differential equations.An importantconsequence the authors would like to draw from this fact is that,contrary to the general belief,in aMarkovian set-up the martingale representation theorem is no longer the reason for the well-posednessof the FBSDE,but rather a consequence of the existence of the solution of the decoupling integralpartialdifferential equation.Finally,the authors briefly discuss the possibility of reducing the regularityrequirements of the coefficients by using a scheme proposed by F.Delarue (2002) to the current case.     

Global convergence of a class of smooth penalty methods for semi-infinite programming

For the semi-infinite programming (SIP) problem, the authors first convert it into an equivalent nonlinear programming problem with only one inequality constraint by using an integral function, and then propose a smooth penalty method based on a class of smooth functions. The main feature of this method is that the global solution of the penalty function is not necessarily solved at each iteration, and under mild assumptions, the method is always feasible and efficient when the evaluation of the integral function is not very expensive. The global convergence property is obtained in the absence of any constraint qualifications, that is, any accumulation point of the sequence generated by the algorithm is the solution of the SIP. Moreover, the authors show a perturbation theorem of the method and obtain several interesting results. Furthermore, the authors show that all iterative points remain feasible after a finite number of iterations under the Mangasarian-Fromovitz constraint qualification. Finally, numerical results are given.     

基于自适应混合能量参数的变步长LMS水声信道均衡算法

提出了一种新的变步长算法,并将该算法用于水声信道均衡。该算法克服改进归一化最小均方(developed normanized least mean square, XENLMS)算法依赖固定能量参数λ的局限性,遵循变步长算法的步长调整原则在XENLMS算法的基础上引入一个自适应混合能量参数λk,改善算法收敛速度和鲁棒性。首先通过仿真分析变步长算法中的3个固定参数α,β,μ的取值范围及对算法收敛性能的影响;并在两种典型的水声信道环境下,采用两种调制信号对算法的收敛性能进行计算机仿真,结果显示,新算法的收敛速度明显快于XENLMS算法和已有的变步长算法,收敛性能接近递归最小二乘(recursive least square, RLS) 算法的最优性能,但计算复杂度远小于RLS算法。最后,木兰湖试验验证了带判决反馈均衡器(decision feedback equalization, DFE)结构的新算法具有较好的克服多径效应和多普勒频移补偿的能力,相比LMS-DFE提高了一个数量级。     

Asymptotic Optimality of the Nonnegative Garrote Estimator Under Heteroscedastic Errors

This paper proposes the Nonnegative Garrote(NG) estimator for linear model with heteroscedastic errors. On the other hand, under some regularity conditions, the authors show the asymptotic optimality of the NG estimator by referring to the idea of the asymptotic optimality of the model average estimator. Simulation results and a real data analysis are reported for testing the results obtained previously. These results provide a stronger theoretical basis for the use of NG estimator by strengthening existing findings.     

Synchronous Parallel Block Coordinate Descent Method for Nonsmooth Convex Function Minimization

This paper proposes a synchronous parallel block coordinate descent algorithm for minimizing a composite function, which consists of a smooth convex function plus a non-smooth but separable convex function. Due to the generalization of the proposed method, some existing synchronous parallel algorithms can be considered as special cases. To tackle high dimensional problems, the authors further develop a randomized variant, which randomly update some blocks of coordinates at each round of computation. Both proposed parallel algorithms are proven to have sub-linear convergence rate under rather mild assumptions. The numerical experiments on solving the large scale regularized logistic regression with 1 norm penalty show that the implementation is quite efficient. The authors conclude with explanation on the observed experimental results and discussion on the potential improvements.     

Approximate conditional likelihood for generalized linear models with general missing data mechanism

The generalized linear model is an indispensable tool for analyzing non-Gaussian response data, with both canonical and non-canonical link functions comprehensively used. When missing values are present, many existing methods in the literature heavily depend on an unverifiable assumption of the missing data mechanism, and they fail when the assumption is violated. This paper proposes a missing data mechanism that is as generally applicable as possible, which includes both ignorable and nonignorable missing data cases, as well as both scenarios of missing values in response and covariate. Under this general missing data mechanism, the authors adopt an approximate conditional likelihood method to estimate unknown parameters. The authors rigorously establish the regularity conditions under which the unknown parameters are identifiable under the approximate conditional likelihood approach. For parameters that are identifiable, the authors prove the asymptotic normality of the estimators obtained by maximizing the approximate conditional likelihood. Some simulation studies are conducted to evaluate finite sample performance of the proposed estimators as well as estimators from some existing methods. Finally, the authors present a biomarker analysis in prostate cancer study to illustrate the proposed method.     

An Efficient Multiple Imputation Approach for Estimating Equations with Response Missing at Random and High-Dimensional Covariates

Empirical-likelihood-based inference for parameters defined by the general estimating equations of Qin and Lawless(1994) remains an active research topic. When the response is missing at random(MAR) and the dimension of covariate is not low, the authors propose a two-stage estimation procedure by using the dimension-reduced kernel estimators in conjunction with an unbiased estimating function based on augmented inverse probability weighting and multiple imputation(AIPW-MI) methods. The authors show that the resulting estimator achieves consistency and asymptotic normality. In addition, the corresponding empirical likelihood ratio statistics asymptotically follow central chi-square distributions when evaluated at the true parameter. The finite-sample performance of the proposed estimator is studied through simulation, and an application to HIV-CD4 data set is also presented.