A Survey on Unsupervised Transfer Clustering

Clustering is widely used in analysis, natural language processing, image segmentation and other data mining fields. However, traditional clustering algorithms, such as K-means, can produce good clustering result only when the instance size is large enough. When the instance size is insufficient, the clustering result will be poor. One way to solve this problem is transfer learning. At present, researches on transfer learning mainly focus on classification and recognition, while researches on clustering are very limited, but become more and more promising. This survey focuses on categorizing and reviewing the current progress on unsupervised transfer clustering algorithm. We also explore some potential future issues in unsupervised transfer clustering research.