Improved fuzzy c-means algorithm based on minimum of distance cost function
Wang Xiaoyun, Lei Shujun · 2011
The traditional fuzzy c-means (FCM) operates when cluster number c is assigned. The value of c makes a great influence on the cluster result. However, the value of cluster number can not be confirmed automatically and needs to be inputted manually, which results in hinders when using the fuzzy c-means. Some researchers have investigated the problem. By combining the concept of distance cost function with the character of fuzzy c-means, this paper improves the FCM algorithm based on new formula of distance cost function. According to calculation of the minimum of modified formula, the optimal cluster number c can be confirmed. The analysis of synthetic and real-world data demonstrate that, improved FCM based on minimum of distance cost function can reach the optimal cluster number.