论文部分内容阅读
提出一种新的结合了模糊c-均值聚类(FCM)算法和可能性c-均值聚类(PCM)算法优点的联合模糊c-均值聚类(AFCM)算法。它克服了PCM对初始值敏感、易产生一致性聚类的缺点,是PCM的扩展算法。试验表明AFCM能同时产生隶属度和典型值,从而更好地处理噪声,避免了一致性聚类,同时提高了聚类准确性。
A new combined fuzzy c-means clustering (AFCM) algorithm that combines the advantages of fuzzy c-means clustering (FCM) algorithm and probabilistic c-mean clustering (PCM) algorithm is proposed. It overcomes the PCM is sensitive to the initial value, easy to produce consistent clustering shortcomings, PCM expansion algorithm. Experiments show that AFCM can generate membership and typical values at the same time, so as to better deal with noise, avoid consistent clustering and improve clustering accuracy.