Transfer Learning for Entropy-Weighted Fuzzy Clustering

The traditional clustering algorithms can not effectively deal with the clustering when the data for current task are not enough. In this paper, we utilize transfer learning to assist the entropy-weighted fuzzy c-means clustering. The clustering centers and corresponding weights of dimensions learned from the known data domain are used in the new objective function to assist the unknown data clustering. Experiments on synthetic data sets have demonstrated the superiority of the new algorithm.