Mini-batch learning of exponential family finite mixture models

Mini-batch algorithms have become increasingly popular due to the requirement for solving optimization problems, based on large-scale data sets. Using an existing online expectation–maximization (EM) algorithm framework, we demonstrate how mini-batch (MB) algorithms may be constructed, and propose a scheme for the stochastic stabilization of the constructed mini-batch algorithms. Theoretical results regarding the convergence of the mini-batch EM algorithms are presented. We then demonstrate how the mini-batch framework may be applied to conduct maximum likelihood (ML) estimation of mixtures of exponential family distributions, with emphasis on ML estimation for mixtures of normal distributions. Via a simulation study, we demonstrate that the mini-batch algorithm for mixtures of normal distributions can outperform the standard EM algorithm. Further evidence of the performance of the mini-batch framework is provided via an application to the famous MNIST data set.     

Relationships among JIT practices: an interpretive modeling approach

Abstract

This research develops a model of relationships among components of Total-JIT, including JIT-information, JIT-manufacturing, JIT-purchasing, and JIT-selling, to establish an implementation hierarchy based on relative importance. The data collected relates to the relationships among JIT components and two performance measures, supply chain competency and organizational performance. Two groups are used in the research, one group of five operations management academics and another group of 30 practicing operations managers working in U.S. manufacturing firms. An interpretive structural modelling methodology is used to develop alternative structural models. The academics’ data show JIT-information emerging as lynchpin of relationships, directly impacting all other JIT practices and both performance measures. The practitioners’ data indicates that all JIT practices and performance measures are interactive as components and outcomes. This study is the first to apply interpretive structural modelling to investigate the interplay among total-JIT components and the performance measures of supply chain competency and organizational performance.     

Network Embeddedness in Exploration and Exploitation of Joint R&D Projects: A Structural Approach

This paper analyses how network embeddedness affects the exploration and exploitation of R&D project performance. By developing joint projects, partners and projects are linked to one another and form a network that generates social capital. We examine how the location, which determines the access to information and knowledge within a network of relationships, affects the performance of projects. We consider this question in the setup of exploration and exploitation projects, using a database built from an EU framework. We find that each of the structural embeddedness dimensions (degree, betweenness and eigenvector centrality) have a different impact on the exploration and exploitation project performance. Our empirical analysis extends to project management literature and social capital theory, by including the effect that the acquisition of external knowledge has on the performance of the project.     

Servant Leadership in the Public Sector: Employee Perspective

Public Organization Review - The purpose of this study is to explore how servant leadership affects public sector employee engagement, organisational ethical climate, and public sector reform, of...     

贫困地区新生代女性农民工休闲生活现状调查研究--以甘肃省定西市为例

休闲已成为现代人生活、工作所不可或缺的组成部分。作为经济转型时期的特殊社会群体,新生代女性农民工的休闲生活现状值得关注。研究以甘肃省定西市为例,从休闲时间、休闲消费、休闲认知三个维度对本地区新生代女性农民工与男性农民工、同龄城市女青年休闲生活进行对比研究,得出结论:贫困地区新生代女性农民工休闲时间严重不足,主要休闲消费方式仍处于消极型的休闲层次类型;相对于上一代农民工而言,她们已经认识到休闲的重要性,但由于受封闭的休闲空间及拮据的经济条件所限,她们的休闲动机仍停留在马斯洛的五个需求层次中前两层的初级需要。针对此现状,文章提出了相应的对策与建议。      

Model averaging inconcentration–QT analyses

This article describes how a frequentist model averaging approach can be used for concentration–QT analyses in the context of thorough QTc studies. Based on simulations, we have concluded that starting from three candidate model families (linear, exponential, and Emax) the model averaging approach leads to treatment effect estimates that are quite robust with respect to the control of the type I error in nearly all simulated scenarios; in particular, with the model averaging approach, the type I error appears less sensitive to model misspecification than the widely used linear model. We noticed also few differences in terms of performance between the model averaging approach and the more classical model selection approach, but we believe that, despite both can be recommended in practice, the model averaging approach can be more appealing because of some deficiencies of model selection approach pointed out in the literature. We think that a model averaging or model selection approach should be systematically considered for conducting concentration–QT analyses. Copyright © 2016 John Wiley & Sons, Ltd.     

Testing Hypotheses in Nonparametric Models of Production

Data envelopment analysis (DEA) and free disposal hull (FDH) estimators are widely used to estimate efficiency of production. Practitioners use DEA estimators far more frequently than FDH estimators, implicitly assuming that production sets are convex. Moreover, use of the constant returns to scale (CRS) version of the DEA estimator requires an assumption of CRS. Although bootstrap methods have been developed for making inference about the efficiencies of individual units, until now no methods exist for making consistent inference about differences in mean efficiency across groups of producers or for testing hypotheses about model structure such as returns to scale or convexity of the production set. We use central limit theorem results from our previous work to develop additional theoretical results permitting consistent tests of model structure and provide Monte Carlo evidence on the performance of the tests in terms of size and power. In addition, the variable returns to scale version of the DEA estimator is proved to attain the faster convergence rate of the CRS-DEA estimator under CRS. Using a sample of U.S. commercial banks, we test and reject convexity of the production set, calling into question results from numerous banking studies that have imposed convexity assumptions. Supplementary materials for this article are available online.     

Finie mixures of canonical fundamenal skew $$$$-disribuions

This paper introduces a finite mixture of canonical fundamental skew \(t\) (CFUST) distributions for a model-based approach to clustering where the clusters are asymmetric and possibly long-tailed (in: Lee and McLachlan, arXiv:1401.8182 [statME], 2014b). The family of CFUST distributions includes the restricted multivariate skew \(t\) and unrestricted multivariate skew \(t\) distributions as special cases. In recent years, a few versions of the multivariate skew \(t\) (MST) mixture model have been put forward, together with various EM-type algorithms for parameter estimation. These formulations adopted either a restricted or unrestricted characterization for their MST densities. In this paper, we examine a natural generalization of these developments, employing the CFUST distribution as the parametric family for the component distributions, and point out that the restricted and unrestricted characterizations can be unified under this general formulation. We show that an exact implementation of the EM algorithm can be achieved for the CFUST distribution and mixtures of this distribution, and present some new analytical results for a conditional expectation involved in the E-step.     

Variable selection for survival data with a class of adaptive elastic net techniques

The accelerated failure time (AFT) models have proved useful in many contexts, though heavy censoring (as for example in cancer survival) and high dimensionality (as for example in microarray data) cause difficulties for model fitting and model selection. We propose new approaches to variable selection for censored data, based on AFT models optimized using regularized weighted least squares. The regularized technique uses a mixture of \(\ell _1\) and \(\ell _2\) norm penalties under two proposed elastic net type approaches. One is the adaptive elastic net and the other is weighted elastic net. The approaches extend the original approaches proposed by Ghosh (Adaptive elastic net: an improvement of elastic net to achieve oracle properties, Technical Reports 2007) and Hong and Zhang (Math Model Nat Phenom 5(3):115–133 2010), respectively. We also extend the two proposed approaches by adding censoring observations as constraints into their model optimization frameworks. The approaches are evaluated on microarray and by simulation. We compare the performance of these approaches with six other variable selection techniques-three are generally used for censored data and the other three are correlation-based greedy methods used for high-dimensional data.