A Study on Sample-Weighted Fuzzy Clustering with Regularizations

• Main Authors: Yi-Shan Pan, 潘儀珊
• Other Authors: Miin-Shen Yang
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/13195164117420389424

碩士 === 中原大學 === 應用數學研究所 === 100 === Abstract In fuzzy cluster analysis, the fuzzy c-means (FCM) clustering algorithm is the most well-known and used method. Up to date, there have been many researches on generalized types of FCM. In 2007, Yu and Yang [25] proposed a generalized fuzzy clustering regularization (GFCR) that adding a regularization with membership functions on an objective function. Besides, some authors tried to solve the affects of the clustering results by outliers and noises. Recently, Yu et al. [4] proposed sample-weighted clustering methods that apply the maximum entropy principle to automatically compute these sample weights for clustering to improve their clustering strengths. In this thesis, we give a sample-weighted version of generalized fuzzy clustering regularization (GFCR), called the sample-weighted fuzzy clustering with regularizations (SW-GFCR). Some numerical examples are considered. These experimental results and comparisons actually demonstrate that the proposed SW-GFCR is more effective and robust than GFCR、SW-KM、SW-FCM and SW-EM.