关键词: 深度子空间聚类, 自表示学习, 指标学习, 亲和矩阵, 正交约束

Abstract:

 Deep subspace clustering is a method that performs well in processing high-dimensional data clustering tasks. However, in the context of challenging data, current deep subspace clustering methods usually exhibit suboptimal clustering results due to the conventional practice of treating self-representative learning and indicator learning as two separate and independent processes. To solve the above problem, a unified enhanced deep subspace clustering method was proposed. Firstly, by integrating feature learning, self-representative learning, and indicator learning together to optimize all parameters, the self-representative matrix was dynamically learned based on the characteristics of the data, ensuring accurate capture of data features. Secondly, to improve the effectiveness of self-representative learning, class prototype pseudo-labels learning was proposed to provide self-supervised information for feature learning and indicator learning, thereby promoting self-representative learning. Finally, to enhance the discriminative ability of embedded representations, orthogonality constraints were introduced to help achieve self-representative attribute. The experimental results show that compared with AASSC (Adaptive Attribute and Structure Subspace Clustering Network), the proposed method improves clustering accuracy by 1.84 percentage points on the MNIST dataset. It can be seen that the proposed algorithm improves the accuracy of self-representative matrix learning, thereby achieving better clustering performance.

Key words: deep subspace clustering, self-representative learning, indicator learning, affinity matrix, orthogonality constraint