Theory and Methods for Estimation of Nonsmooth Functionals and Detection of Simultaneous Signals

非光滑泛函估计和同时信号检测的理论和方法

• 批准号: 1403708
• 负责人: T. Tony Cai
• 金额: $ 48.58万
• 依托单位: University of Pennsylvania
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-07-01 至 2017-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1403708&HistoricalAwards=false
• 关键词: Theory; Methods; Estimation; Nonsmooth; Functionals

The collection and analysis of high dimensional data is now of central importance in a wide range of scientific applications, including genomics, bioinformatics, climate studies, and signal processing. There are many new challenges in the analysis of such high dimensional data and there is much demand for the development of new statistical theory and methodology in this field. Although high dimensional data is often analyzed using complex models in many applications, attention is often focused on testing specific hypotheses that are of low dimension. In such cases new statistical theory and methodology relies on the development of new techniques for estimating non-smooth functionals, the topic of this project.The focus of this research project is to establish a comprehensive methodology for estimating non-smooth functionals and to show how this methodology can be used in the detection and testing of multiple high-dimensional sparse sequences encountered in genomics. The proposed research will provide scientists with new methodologies to analyze such highly complex data. The statistical procedures developed in this project will be implemented in a high-level programming language and made available on the internet with the related research reports so as to allow comparisons with other approaches.

高维数据的收集和分析现在在广泛的科学应用中具有核心重要性，包括基因组学，生物信息学，气候研究和信号处理。这种高维数据的分析面临着许多新的挑战，需要发展新的统计理论和方法。虽然在许多应用中，高维数据经常使用复杂的模型进行分析，但人们的注意力往往集中在测试低维的特定假设上。在这种情况下，新的统计理论和方法依赖于估计非光滑泛函的新技术的发展，这是本项目的主题。本研究项目的重点是建立一种综合的方法来估计非光滑函数，并展示如何将这种方法用于基因组学中遇到的多个高维稀疏序列的检测和测试。拟议中的研究将为科学家们分析如此高度复杂的数据提供新的方法。本项目所制订的统计程序将以一种高级编程语言执行，并与有关的研究报告一起在互联网上提供，以便与其他方法进行比较。