查詢結果分析
來源資料
相關文獻
- A Tabu Search-based Algorithm for the Cluster Validity Problems
- An Automatic Clustering Approach to Classifying Auditory Neurons
- An Improved Tabu Search for Vector Quantization
- 臺灣南、北地區紅樹林區魚種組成之比較
- 使用尼曼A型分配於積體電路晶圓缺陷管制圖之經濟性設計
- 金門農業試驗所鳥類群聚的組成與結構
- 模糊資料之聚類分析
- 墾丁國家公園南灣海域珊瑚群聚的變遷:1987至1997年
- Spatial and Temporal Variations of Fish Community in the Mangrove Swamps Along the Western Coast of Taiwan
- 關刀溪森林生態系兩棲動物群聚之研究
頁籤選單縮合
題名 | A Tabu Search-based Algorithm for the Cluster Validity Problems |
---|---|
作者姓名(中文) | 潘世明; | 書刊名 | 高苑學報 |
卷期 | 13 2007.07[民96.07] |
頁次 | 頁67-90 |
分類號 | 312.2 |
關鍵詞 | 群聚; 群聚正確性; 演化運算法; 禁制搜尋法; Clustering; Cluster validity; Evolutionary algorithms; Tabu search; |
語文 | 英文(English) |
中文摘要 | 對於群聚之正確性問題的最常見解法是在選擇一適當之群聚正確性指標參數後,再依序指定每一可能的群聚個數給所選擇之傳統群聚演算法來建構各種可能之群聚結構,最後從中選出具最佳正確性指標值之群聚結構。此種解法需要一個冗長的過程。為了能更有效率地解群聚之正確性問題,本研究以禁制搜尋法之概念為基礎提出一種自動群聚演算法。此演算法不需使用者依序提供每一個可能的群聚個數,而能夠自動決定出最佳群聚個數且同時建構出具最佳正確性指標值之群聚結構。在本演算法中,群聚個數被視為一個變數且最終演化成一個最佳值。在實驗中,三個人造資料集及與三個真實資料集用來測試本演算法的性能。結果證實本演算法優於五種傳統解法。 |
英文摘要 | For the cluster validity problems, the most commonly used method is that one se-lects or defines a cluster validity index and performs a traditional clustering algorithm for all possible numbers of clusters in sequence to find the clustering structure with the best cluster validity. This process is a tedious and time-consuming work. To efficiently solve the cluster validity problems (i.e., determining the optimal number of clusters and constructing the clusters with good validity at the same time), an automatic clustering algorithm that does not require users to give each possible number of clusters is pro-posed based on the concepts of tabu search. The proposed algorithm treats the number of clusters as a variable and evolves it to an optimal number. Compared with five types of exhaustive searches using different traditional clustering algorithms, the superiority of the proposed algorithm in solving the cluster validity problems is demonstrated through experiments conducted on three artificial data sets and three real-life data sets. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。