Abstract

Fuzzy entropy clustering is an improved fuzzy C-means algorithm and is proposed in the past years. In this paper, by introducing the generalized entropy into fuzzy clustering, we obtain the objective function of the generalized entropy, and use neural networks and the augmented Lagrange method to solve the optimization problem with objective function of generalized entropy. Afterwards, we present a generalized entropy's fuzzy clustering algorithm based on neural network, namely GEFCMNN. Meanwhile, we use two ways to assign values of the augmented Lagrange multipliers which are randomly selected method and iterative method, respectively. In the experiment, we selected three different data sets to test GEFCMNN algorithm's performance. It can be seen from experimental results that when weighted index m is greater than 2, better clustering results can be obtained.

Paper Details