2009 Workshop on Semiparametric methodology

半参数方法仍然是一个活跃的研究领域，特别是随着计算资源和能力的增长。这些方法用于参数和非参数分量模型的推断。工作范围从半参数建模和效率到半参数回归。前者涉及在无限维干扰参数的存在下进行有效的推断。后者对应于具有参数误差和未指定（非参数）回归函数的回归模型。在高维缺失数据问题中，无论是在频率论和贝叶斯推理的背景下，还是在估计回归函数中，基础工作仍有待完成，这些问题在许多不同的应用领域中都存在复杂性。该研讨会汇集了半参数方法学领域的领导者和年轻的研究人员。被邀请的演讲者都是非常杰出的个人，他们在半参数学方面做了杰出的工作，并共同撰写了三篇论文。具有较少假设的灵活建模是解决许多科学问题不可或缺的组成部分。半参数方法是实现这一点的一种方法。在过去的30年里，回归分析已经取得了很大的进展，但是在不完全数据的高维问题、回归函数的估计以及其他一些应用领域中的复杂性等方面，仍有一些基础性的工作要做。这些工作的应用范围很广，包括时空模型，捕获-再捕获模型，环境应用中的测量误差模型，遗传学和基因组学中的分类和测试，以及医学观察研究和临床试验中的缺失数据模型。研讨会提供了一个极好的机会来讨论半参数方法的许多最新的重大发展，并确定重要的问题和新的研究方向。

Semiparametric methods continue to be an active research area,especially as computing resources and power grow. These methods are usedfor inference in models with both parametric and nonparametriccomponents. Work in this setting ranges from semiparametric modeling andefficiency to semiparametric regression. The former involves makingefficient inferences in the presence of infinite dimensional nuisanceparameters. The latter corresponds to regression models with parametricerrors and unspecified (nonparametric) regression functions. Fundamentalwork remains to be done in high dimensional missing data problems, bothin the setting of frequentist and Bayesian inference and in estimatingregression functions with the complications posed in many differentareas of application. The workshop brings together leaders in the fieldof semiparametric methodology and young researchers. The invitedspeakers all are very distinguished individuals who have doneoutstanding work in semiparametrics and have co-authored three of theseminal texts. Flexible modeling with few assumptions is an integral component ofsolving many scientific problems. Semiparametric methods are a way toaccomplish this. Major advances have been made in the past thirty years.However, fundamental work remains to be done in high dimensionalproblems with incomplete data, in estimating regression functions withthe complications posed in many different areas of application, amongother areas. Applications for such work are wide ranging and includespatio-temporal models, capture-recapture models, and measurement errormodels in environmental applications, classification and testing in ingenetics and genomics, and missing data models in longitudinalobservational studies and clinical trials. The workshop provides anexcellent opportunity to discuss the many recent significantdevelopments in semiparametric methodology and to identify importantproblems and new research directions.