Flexible Statistical Modeling

研究人员研究了各种应用情况下的统计模型，这些情况需要对标准技术进行创新修改。文档分类在极高维的特征空间中建立模型，就像用基因表达阵列进行推理和预测的模型一样。物种发生和丰度模型涉及大量物种，通常具有许多共同的特征。在每一种情况下，不同的背景都导致研究人员开发出特殊的正则化形式，允许人们利用数据中的结构。在这个先进的技术时代，我们面临着分析极其大量的数据。这位研究人员涉及的几个例子中有两个是基因表达测量(每个人类样本4万个测量)和在线文档分类(通常网络是来源)。标准统计工具在这些情况下不能很好地发挥作用。这位研究人员研究了这些工具的创新适应，旨在克服这些问题带来的挑战。

The investigator studies statistical models in a variety of appliedsituations which require innovative modifications of the standardtechnology. Document classification builds models in extremelyhigh-dimensional feature spaces, as do models for inference andprediction with gene expression arrays. Species occurrence andabundance models deal with large numbers of species, often sharingmany characteristics. In each of these settings, the differentcontexts have led the researcher to develop special forms ofregularization that allow one to exploit the structure in the data.In this advanced technological age, we are faced with analyzing extremely largevolumes of data. Two of the several examples this researcher deals with are geneexpression measurements (40 thousand measurements per human sample), andonline document classification (often the web is the source). Standardstatistical tools do not work well in these situations. This investigatorstudies innovative adaptations of these tools designed to defeat thechallenges these problems pose.