The Fuzzy Topological Map (FTM) algorithm is a topological clustering algorithm modified from Kohonen Feature Map (KFM) algorithm. In particular, the spring-like attraction law is replaced with an progressive attraction law which tunes from magnet-like to spring-like attraction during training. The magnitude of the attraction is obtained in a similar way used in the Fuzzy c-Mean (FCM) algorithm. With modified attraction law, the FTM algorithm is similar to a sequential FCM algorithm at the beginning, and to KFM algorithm at the end. The modified attraction law enables a FTM algorithm to exhibit speedier convergent rates and more regularity convergent patterns relative to that of KFM algorithm. In this paper, we will present the FTM algorithm, analysis on convergence, and the results on...[ Read more ]

The Fuzzy Topological Map (FTM) algorithm is a topological clustering algorithm modified from Kohonen Feature Map (KFM) algorithm. In particular, the spring-like attraction law is replaced with an progressive attraction law which tunes from magnet-like to spring-like attraction during training. The magnitude of the attraction is obtained in a similar way used in the Fuzzy c-Mean (FCM) algorithm. With modified attraction law, the FTM algorithm is similar to a sequential FCM algorithm at the beginning, and to KFM algorithm at the end. The modified attraction law enables a FTM algorithm to exhibit speedier convergent rates and more regularity convergent patterns relative to that of KFM algorithm. In this paper, we will present the FTM algorithm, analysis on convergence, and the results on a number of well-known cases.