Self-Expressive Based Multi-View Clustering Via Contrastive Learning

In the past decade, deep learning has made great progress in Multi-view Clustering (MvC). MvC utilizes deep neural networks to explore and exploit complex relationships among multiple views in an unsupervised manner, aiming to learn compact data clusters. Numerous studies have utilized diverse loss functions to direct the learning process of the network, including the minimization of reconstruction loss. Contrastive learning has been applied to train models on unlabeled data, which can improve the performance and generalization ability of the models. The objective of contrastive learning is to minimize the distance between samples of the same class while maximizing the distance between samples of different classes. The core idea behind contrastive learning is to learn feature representations by leveraging the inherent structure of the data. In this paper, we introduce a model named self-expressive based multi-view clustering via contrastive learning(SMC-CL) . The first loss, referred to as the contrastive loss, extracts individual sample features from multiple views considering the relationships between samples. The second loss is used to measure the effectiveness of the fused features from different views.