中大機構典藏-NCU Institutional Repository-提供博碩士論文、考古題、期刊論文、研究計畫等下載:Item 987654321/82285
English  |  正體中文  |  简体中文  |  全文笔数/总笔数 : 78728/78728 (100%)
造访人次 : 33565907      在线人数 : 649
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜寻范围 查询小技巧:
  • 您可在西文检索词汇前后加上"双引号",以获取较精准的检索结果
  • 若欲以作者姓名搜寻,建议至进阶搜寻限定作者字段,可获得较完整数据
  • 进阶搜寻


    jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/82285


    题名: 高維度電腦實驗設計:智能篩選與訊息擷取( I );High-Dimensional Computer Experiments: Intelligent Screening and Informative Sampling( I )
    作者: 張明中
    贡献者: 國立中央大學統計研究所
    关键词: 大數據;混沌多項式;因子篩選;空間填充;資訊子資料;Big data;Polynomial chaos;Factor screening;Space-filling;informative subdata
    日期: 2020-01-13
    上传时间: 2020-01-13 14:36:43 (UTC+8)
    出版者: 科技部
    摘要: ``高維度''一詞充斥在許多新的研究領域。以往該詞意指變數個數遠大過資料量。近來由於科技進步,數據常有資料量遠大過變數個數的情況,因此高維度可泛指前述任一情境。本計畫致力在高維度電腦實驗下,發展高品質的因子篩選設計,與準確的估計參數/預測反應曲面的方法,完整的理論結構與快速運行的程式套件將是本計畫執行重點之一。在計畫的第一部分,我們將發展當p>>n時,可快速篩選出重要因子之方法論。此方法論預計可容易執行,且不須太耗費硬體資源。在計畫的第二部分,我們將探討當n>>p時,利用混沌多項式來估計電腦實驗的模型參數與預測反應曲面。除了完整的理論價構外,有效率的演算法也是發展重點。本計畫的成果預計可改善傳統方法的效能,並期對社會產業有所貢獻。 ;With technological advances, the terminology “high-dimensional” is much more commonly seen nowadays than a few decades ago. It used to refer to “the number of variables are much greater than the sample size” (i.e., p>>n). Recently, there has been an exponential growth in the size of datasets. Therefore, high-dimensional data may also refer to those whose data sizes are much greater than the numbers of variables (i.e., n>>p). This project aims at developing an efficient screening methodology and highly accurate estimation/prediction techniques for high-dimensional computer experiments. In the first part, we plan to develop a new methodology that can effectively identify active factors, with linear/nonlinear effects, for the case of p>>n. This methodology is expected to perform well and easily implemented with little computational burden. In the second part, we study subdata selection through polynomial chaos expansions when n>>p. The proposed methods would cover several common scenarios. Theoretical properties and efficient algorithms are to be provided. To the best of our knowledge, all relevant research to date, for either p>>n or n>>p, has yet to be fully satisfactory. This project tries to fulfill these gaps and devotes to the theory and application that can be easily applied to real-world situations.
    關聯: 財團法人國家實驗研究院科技政策研究與資訊中心
    显示于类别:[統計研究所] 研究計畫

    文件中的档案:

    档案 描述 大小格式浏览次数
    index.html0KbHTML319检视/开启


    在NCUIR中所有的数据项都受到原著作权保护.

    社群 sharing

    ::: Copyright National Central University. | 國立中央大學圖書館版權所有 | 收藏本站 | 設為首頁 | 最佳瀏覽畫面: 1024*768 | 建站日期:8-24-2009 :::
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 隱私權政策聲明