English  |  正體中文  |  简体中文  |  Items with full text/Total items : 49378/84106 (59%)
Visitors : 7360702      Online Users : 53
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    Please use this identifier to cite or link to this item: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/46972


    Title: 支援向量模糊灰色預測系統之建構與分析(I)
    Other Titles: Support Vector Fuzzy Grey Forecasting System (I)
    Authors: 曹銳勤
    Contributors: 淡江大學經營決策學系
    Date: 2009
    Issue Date: 2010-04-15 15:46:18 (UTC+8)
    Abstract: 支援向量機(SVM)在圖形辨識上已獲得非常顯著的成效。而且, 在 Vapnik 提出ε -感覺遲鈍損失函數(ε -insensitive loss function) 後,SVM 已 經拓展至解決線性及非線性迴歸的估計, 而此方法被命名為支援向量迴歸 (SVR)。爾後, Hong & Hwang 拓展SVR 模式以處理變數間具模糊關係或 所蒐集的資料為模糊資料, 而提出支援向量多元模糊線性和非線性迴歸模 式以有效解決具模糊性的問題, 使支援向量多元模糊迴歸模式的基礎更完 整並提升計算效率。爾後學者在支援向量模糊迴歸機的研究過程中分別針 對非對稱的支援向量機的研究、討論在非線性的迴歸系統中 hybrid kernel 對模式求解的影響、用類神經網路混合求解非線性支援向量模糊迴歸的參 數等, 諸多研究對於模式的建構與預測績效的提升皆有顯著的助益。然而 至目前為止, 上述所有的研究皆是不確定性的因果分析問題, 對於使用支 援向量模糊迴歸模式解決稀少性或模糊性的時間序列資料,並無任何討論。 因此, 在本計畫的第一年研究中, 我們計畫建構支援向量(模糊)灰色 預測系統以解決稀少性或模糊性的時間序列數據。在此預測系統中包括支 援向量灰色模式GM(1,1)、支援向量模糊灰色模式GM(1,1)、支援向量灰色 模式GM(1,N)及支援向量模糊灰色模式GM(1,N)。運用ε -感覺遲鈍損失函 數(ε -insensitive loss function), 大多數被收集的時間序列資料將被歸入 ε -tube 中, 由於只有落在ε -tube 外的資料對於模式的建構與參數的求解 具影響性, 因此預測績效不會隨著資料量愈來愈多而愈來愈差, 這是目前 灰預測無法有效解決的問題。在第2 年的研究中, 我們也將提出支援向量 時間序列模式及支援向量模糊時間序列模式, 在此計畫的模式中本研究將 提出較以低資料量分析時間序列資料的模式, 以解決統計時間序列模式需 要大量資料建構模式的限制; 而支援向量模糊時間序列資料亦將同時被討 論。此外, 本研究對於提出的預測系統亦將進行敏感性分析以探討本研究 所提出的預測系統具一致的預測能力。最後, 本研究將舉各個不同產業的 需求分析預測及趨勢預測以說明本研究的可應用性。 Support vector machine (SVM) has been very successful in pattern recognition. Moreover, introduction of ε- insensitive loss function by Vapnik, SVM has been developed in solving non-linear regression estimation, a new techniques called support vector regression (SVR). Then, Hong & Hwang proposed SVM for multivariate fuzzy linear and nonlinear regression models. Using the basic idea underlying SVR for multivariate fuzzy regressions gives computational efficiency of getting solutions. In the support vector fuzzy regression machine, for solving crisp-input-fuzzy-output and crisp-input-fuzzy-output data, some researches such as asymmetric support vector machine, hybrid kernel research in nonlinear regression system, a hybrid model using neural network for solving support vector nonlinear fuzzy regression machine are the interesting topics by the followers. However those of above researches are all of cause-effect analysis, we do not know how to analyze time series data using the support vector fuzzy regression machine. In this project of the first year, we first propose a support vector (fuzzy) grey forecasting system for solving time series data, including support vector grey model GM(1,1), support vector fuzzy grey model GM(1,1), support vector grey model GM(1,N), and support vector fuzzy grey model GM(1,N). By using the ε-insensitive functions for function estimation in the proposed models, most of the collected data should be included in an ε-tube of accuracy, and then more precise forecasting error could be derived whereas the forecasting error of grey model is worse as the collected is increasing. In the second year, we also will propose both of support vector time series model and support vector fuzzy time series model which are never considered before. Because the proposed approach can provide a better initial structure for analyzing time series data, it can have a more precise forecasting ability. Some sensitivity analysis will be used to verify the consistent forecasting ability in the proposed forecasting system. Finally, some applications will be illustrated to forecast the demand and trend of some industries.
    Appears in Collections:[管理科學學系暨研究所] 研究報告

    Files in This Item:

    There are no files associated with this item.

    All items in 機構典藏 are protected by copyright, with all rights reserved.


    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - Feedback