深度聚类研究综述OA

Clustering plays an important role in machine learning,and learning a good representation of data is essential for clustering.Deep clustering joint optimization representation learning and clustering models have been widely used in various clustering tasks.The paper overviews the recent advances in deep clustering and summarize its applications in different fields.Firstly,the basic concepts and principles of deep clustering are introduced,and from the perspective of neural network model,it is divided into autoencoder-based,variational-autoencoder-based,generative adversarial network-based,twin-based network/contrastive learning,and deep clustering based on graph neural network,which are analyzed and summarized respectively.Then,the applications of deep clustering in the fields of image,text,and recognition detection are discussed.Finally,the current hotspots and future development directions of deep clustering research are prospected.