區間資料之特徵簡化模糊C均值演算法

A Feature-Reduction Fuzzy C-Means Algorithm for Interval Data

劉宛儒(Wan-Ru Liu)

指導教授 ： 楊敏生

中原大學/理學院/應用數學研究所/碩士(2018年)

https://doi.org/10.6840/cycu201800411

摘要

模糊聚類演算法通常將每筆資料的每一個特徵分量的資訊都視為同樣重要，在大量的資料中往往會出現不重要的訊息，然而在演算法過程中涉及資料裡不重要或不相關的訊息，可能會導致演算法的效果有誤差。最近Yang和Nataliani(2017)所提出了一種改善FCM演算法的方法，其演算法可以自動計算每個訊息的重要性，也就是說要根據訊息的相關性給予不同的重要性，並同時減少不相關的資訊，然而FRFCM演算法只針對一般資料，但現在區間值資料也常被使用，如天氣的最高溫與最低溫、物體的凝固點與沸點等，本文將擴展FRFCM演算法來處理區間值資料，並進一步探討區間值資料對於演算法的有效性和實用性。

關鍵字

聚類演算法 ； 模糊C均值 ； 特徵簡化 ； 區間資料

並列摘要

Fuzzy clustering algorithm usually considers equally important information for all feature components of data. In a large amount of data, unimportant messages may appear for some feature components. However, in the process of most fuzzy clustering algorithms, information are always involved in even though they are not important or relevant in the data. This may cause errors in results of fuzzy clustering algorithms. Recently, Yang and Nataliani (2017) proposed a method to improve the fuzzy c-means (FCM) algorithm, called Feature-Reduction Fuzzy C-Means (FRFCM) algorithm. The FRFCM algorithm can automatically calculate different importance of feature components. That is, it can give different importance according to the relevance of the message, and at the same time it can also reduce the related information. However, the FRFCM algorithm is only for general data, not for interval value data. We know that interval value data are also often used, such as the highest temperature and the lowest temperature of the weather, the freezing point and the boiling point of the object, etc. This thesis will expand the FRFCM algorithm to treat interval value data. We further discuss validity and practicability of the proposed clustering algorithm for interval-valued data.

並列關鍵字

Clustering algorithm ； Fuzzy c-means ； Feature-reduction ； Interval data

參考文獻

[1] L. A.Zadeh. (1965). "Fuzzy Sets," Information and Control, vol. 8, pp. 338-353.

[2] L. Jing, M. K. Ng and J. Z. Huang. (2007). "An Entropy Weighting k-Means Algorithm for Subspace Clustering of High-Dimensional Sparse Data," IEEE Transactions on Knowledge and Data Engineering, vol. 19, pp. 1026-1041.

[3] X. Wang, Y. Wang and L. Wang.(2004) "Improving fuzzy c-means clustering based on feature weight learning," Pattern Recognition Letters, vol. 25, pp. 1123-1132.

[4] M.S. Yang, Y. Nataliani. (2017). "A Feature-Reduction Fuzzy Clustering Algorithm Based on Feature-Weighted Entropy", IEEE Transactions on Fuzzy Systems, vol 26, pp. 817-835.

[5] M.S. Yang, J.H. Yang. (2010). "模糊聚類及其應用", 藍海文化, pp. 18-20.