Regularization Using QR Factorization and the Estimation of the Optimal Parameter

In this paper we propose a direct regularization method using QR factorization for solving linear discrete ill-posed problems. The decomposition of the coefficient matrix requires less computational cost than the singular value decomposition which is usually used for Tikhonov regularization. This method requires a parameter which is similar to the regularization parameter of Tikhonov's method. In order to estimate the optimal parameter, we apply three well-known parameter choice methods for Tikhonov regularization.This revised version was published online in October 2005 with corrections to the Cover Date.     

Almost Optimal Differentiation Using Noisy Data

We study differentiation of functionsfbased on noisy dataf(t_i)+_i. We recoverf^(k)either at a single point or on the interval [0, 1] inL₂-norm. Under stochastic assumptions onfand_i, we determine the order of the errors of the best linear methods which use n noisy function values. Polynomial interpolation for the pointwise problem and smoothing splines for the problem inL₂-norm are shown to be almost optimal. The analysis involves worst case estimates in reproducing kernel Hilbert spaces and a Landau inequality.     

Email Surveillance Using Non-negative Matrix Factorization

In this study, we apply a non-negative matrix factorization approach for the extraction and detection of concepts or topics from electronic mail messages. For the publicly released Enron electronic mail collection, we encode sparse term-by-message matrices and use a low rank non-negative matrix factorization algorithm to preserve natural data non-negativity and avoid subtractive basis vector and encoding interactions present in techniques such as principal component analysis. Results in topic detection and message clustering are discussed in the context of published Enron business practices and activities, and benchmarks addressing the computational complexity of our approach are provided. The resulting basis vectors and matrix projections of this approach can be used to identify and monitor underlying semantic features (topics) and message clusters in a general or high-level way without the need to read individual electronic mail messages. Michael W. Berry is a Professor and Interim Department Head in the Department of Computer Science at the University of Tennessee and a faculty member in the Graduate School in Genome Science and Technology Program at the University of Tennessee and Oak Ridge National Laboratory. His research interests include information retrieval, data mining, scientific computing, computational science, and numerical linear algebra. He is a member of the Society for Industrial and Applied Mathematics (SIAM), Association for Computing Machinery (ACM), and the Computer Society of the Institute of Electrical and Electronics (IEEE). Professor Berry is on the editorial boards of “Computing in Science and Engineering” (IEEE Computer Society and the American Institute of Physics) and the SIAM Journal of Scientific Computing. Murray Browne is a Research Associate in the Department of Computer Science at the University of Tennessee. He is a member of the American Society for Information Science and Technology and has published numerous essays, book reviews, newspaper articles, and feature stories.     

Phylogenetic Inference for Binary Data on Dendograms Using Markov Chain Monte Carlo

Abstract

Using a stochastic model for the evolution of discrete characters among a group of organisms, we derive a Markov chain that simulates a Bayesian posterior distribution on the space of dendograms. A transformation of the tree into a canonical cophenetic matrix form, with distinct entries along its superdiagonal, suggests a simple proposal distribution for selecting candidate trees “close” to the current tree in the chain. We apply the consequent Metropolis algorithm to published restriction site data on nine species of plants. The Markov chain mixes well from random starting trees, generating reproducible estimates and confidence sets for the path of evolution.     

Binary Optimal Odd Formally Self-Dual Codes

In this paper, we study binary optimal odd formallyself-dual codes. All optimal odd formally self-dual codes areclassified for length up to 16. The highest minimum weight ofany odd formally self-dual codes of length up to 24 is determined. We also show that there is a unique linearcode for parameters [16, 8, 5] and [22, 11, 7], up to equivalence.     

Bootstrapping Pseudolikelihood Models for Clustered Binary Data

Asymptotic properties of the parametric bootstrap procedure for maximum pseudolikelihood estimators and hypothesis tests are studied in the general framework of associated populations. The technique is applied to the analysis of toxicological experiments which, based on pseudolikelihood inference for clustered binary data, fits into this framework. It is shown that the bootstrap approximation can be used as an interesting alternative to the classical asymptotic distribution of estimators and test statistics. Finite sample simulations for clustered binary data models confirm the asymptotic theory and indicate some substantial improvements.     

Construction of Pseudorandom Binary Sequences Using Additive Characters

In earlier papers the authors studied finite pseudorandom binary sequences, and they constructed sequences with strong pseudorandom properties. In these earlier constructions multiplicative characters were used. In this paper a new construction is presented which utilizes properties of additive characters. These new sequences can be computed fast, they are well-distributed relative to arithmetic progressions and their correlations of small order are small, but the price paid for the fast computation is that the correlations of large order can be large.     

Construction of Pseudorandom Binary Sequences Using Additive Characters

In earlier papers the authors studied finite pseudorandom binary sequences, and they constructed sequences with strong pseudorandom properties. In these earlier constructions multiplicative characters were used. In this paper a new construction is presented which utilizes properties of additive characters. These new sequences can be computed fast, they are well-distributed relative to arithmetic progressions and their correlations of small order are small, but the price paid for the fast computation is that the correlations of large order can be large.     

多变量系统的基于矩阵$K_{P}=LDU$分解的鲁棒模型参考自适应控制

本文针对具有未建模动态的多变量系统,研究了基于高频增益矩阵KP=LDU分解的鲁棒直接型模型参考自适应控制问题,严格地分析了闭环系统的稳定性和鲁棒性．     

二元选择面板模型的截面相关检验

鉴于CD统计量对截面相关形式是不稳健的,需对CD统计量进行修正。从而本文提出了渐近服从标准正态分布的MCD统计量,与已有的LM_(BC)统计量和CD统计量比较发现:(1)三个统计量不存在明显的水平扭曲;(2)当截面间相关度有正有负,且规模较为接近时,CD统计量的功效很低,MCD统计量和LM_(BC)统计量对于截面相关方式更稳健,而在同向截面相关模型中CD统计量的功效明显更好;(3)当模型存在分布误设或异方差时,CD统计量、LM_(BC)统计量和MCD统计量的表现都较为稳健。我们将这三个统计量应用到国内上市公司现金分红行为模型中,发现公司间分红行为具有显著的相关性。     

基于时序矩阵分解的缺失销售数据估计

企业的历史销售记录是供应链优化研究的基础数据来源,然而,在日常的研究中,几乎所有可以通过公开途径获得的销售记录都是高度不完整的,这为研究者开展工作带来了极大的不便。为解决此问题,本文提出,以销售数据集中已有的数据为基础,使用面向时序数据的矩阵分解模型MAFTIS对其缺失的部分进行估算,从而把残缺的数据集补全完整。进一步地,为提高MAFTIS的计算效率,本文还为该模型设计了一种基于交替最小二乘法的求解策略MAFTIS^ALS。在评估实验中,MAFTIS^ALS被用于三个真实销售数据集的缺失记录估计,结果显示,与其它估计模型相比,MAFTIS^ALS能获得更准确的估计结果,并且具有更高的收敛速度。     

Optimal Recursive Estimation of Raw Data

We present a new approach to the optimal estimation of random vectors. The approach is based on a combination of a specific iterative procedure and the solution of a best approximation problem with a polynomial approximant. We show that the combination of these new techniques allow us to build a computationally effective and flexible estimator. The strict justification of the proposed technique is provided. This work was supported by the Australian Research Council under the ARC Large Grant Scheme.     

利用射影平面构作最优二元等重码

二元等重码(n,M,2δ,ω)在通讯中有着极为广泛的应用.利用m阶射影平面的性质构作了一个二元等距等重码,并且研究了这个二元等重码成为最优等重码的条件,找出了一个最优检错码.     

Factorization of factorials

If factorials enter into product-quotient expressions, not allowing approximative calculation, the factorization into powers of primes is recommended. The paper describes a computer program for this process, based on a known formula, which should be applied differently, according as the primes are large or small. A discussion shows how to distinguish between large and small in order to minimize the computing time.     

成败型数据下Logistic分布的参数估计

使用EM算法 ,在成败型数据下 ,对Logistic分布的参数进行估计 ,得到了估计量所满足的非线性方程组     

Empirical Best Prediction for Small Area Inference with Binary Data

The paper introduces a frequentist's alternative to the recently developed hierarchical Bayes methods for small area estimation with binary data. Specifically, the best predictor (BP) and empirical best predictor (EBP) of small area specific random effect are developed in the context of a mixed logistic model and different asymptotic properties of the proposed BP and EBP are studied. An approximation to the mean squared error (MSE) of the proposed EBP correct up to the order o(m ^–1) is obtained, where m denotes the number of small areas. The asymptotic behavior of the relative savings loss (RSL) demonstrates the superiority of the proposed EBP over the usual small area proportion.