標題: 群聚分析與模糊自適應共振理論之比較研究
A Comparison of Cluster Analysis and Fuzzy ART
作者: 陳琬渝
Chen, Wan-Yu
唐麗英
Lee-Ing Tong
工業工程與管理學系
關鍵字: 聚類;群聚分析;自適應共振理論;模糊自適應共振理論;cluster;cluster analysis;Adaptive Resonance Theory;Fuzzy Adaptive Resonance Theory
公開日期: 1996
摘要: 在學術界與工業界有許多領域常常需要使用到聚類(cluster)的技術 ,來將物件或觀察體分類。多變量統計方法中的群聚分析(Cluster Analysis)是解決聚類問題的傳統方法。而模糊自適應共振理論(Fuzzy AdaptiveResonance Theory,簡稱Fuzzy ART)則是新興的類神經網路( artificialneural network)模式,亦可用來解決聚類問題。本研究的目 的即是比較群聚分析與模糊自適應共振理論在解決聚類問題上的異同點, 並指出兩者各有何優缺點,以供學術界或工業界在選擇聚類技術時參考之 用。本研究並以三個實例來說明此二種方法在實際應用上之差異處及優缺 點﹕一為紫蘿蘭花依花萼長、花萼寬、花瓣長、花瓣寬等變數作分類﹔一 為大學與高中足球員依頭部尺寸來做分類﹔另一則為群組技術(Group Technology)的應用,利用群聚分析與模糊自適應共振理論將工件與機器 分群以形成單元製造(celluar manufacture),再比較此兩種方法在群組 技術上之優劣。本研究的結論為Fuzzy ART在聚類的準確性上,較群聚分 析略優,且執行時間較快,在聚類過程中也不需加入人為的判斷。 In many areas, cluster techniques are often used to classify the subjects that are similar to each other with respective to certain common characteristics. Cluster Analysis is a conventional multivariate statistical technique to form the homogeneous groups or clusters. Fuzzy Adaptive Resonance Theory( Fuzzy ART), one of the new artificial neural networks, is an alternative cluster technique. The objective of this thesis is to compare the Cluster Analysis and Fuzzy ART in solving the clustering problems. Three real-word cases are given in this study to compare the effectiveness of the Cluster Analysis and Fuzzy ART. The results indicate that Fuzzy ART is more efficient and accurate in classifying the subjects into groups.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT850031043
http://hdl.handle.net/11536/61486
顯示於類別:畢業論文