Abstract: | 在許多的情況裡, 尤其是現代高度競爭的環境中, 往往在資料尚不足的情況下, 決策必須快速地完成, 因此, 如何提高小資料學習(small data set learning)的準確度, 成為一個重要的研究主題。 在筆者之前的研究裡(Li, et al., 2005a; Li, et al., 2005b), 提出了應用資料模糊化之總合模糊法(mega-fuzzification), 以解決彈性製造系統中小資料學習的問題。該方法將確定性資料(crisp data)轉化為連續性資料, 同時將連續性資料之資料領域外擴至其真正的資料範圍; 然而, 在很多的情況裡, 資料的真正領域範圍並非是已知的, 因此本研究將進一步探討, 當資料領域是未知的情況下, 如何決定資料外擴的程度。Knowledge derived from limited data gathered in the early system stages is usually too fragile. Unfortunately, decisions have to be made quickly in a competitive environment. In our previous study, a strategy using continuous data and domain external expansion methods under a known data domain range was proposed to solve the so called small data set learning problem in FMS. This research goes further in seeking a quantitative method to determine the range of domain external expansion under unknown domain bounds. This research considers the data bias phenomenon that often occurs in small data sets and provides a method for its adjustment. |