Abstract:The goal of multi-view clustering is to partition samples into different subsets according to their diverse features. Previous multi-view clustering methods mainly exist two forms: multi-view spectral clustering and multi-view matrix factorization. Although they have shown excellent performance in many occasions, there are still many disadvantages. For example, multi-view spectral clustering usually needs to perform postprocessing. Multi-view matrix factorization directly decomposes the original data features. When the size of features is large, it encounters the expensive time consumption to decompose these data features thoroughly. Therefore, we proposed a novel multi-view clustering approach. The main advantages include the following three aspects: 1) it searches for a common joint graph across multiple views, which fully explores the hidden structure information by utilizing the compatibility among views; 2) the introduced nonnegative constraint manipulates that the final clustering results can be directly obtained; and 3) straightforwardly decomposing the similarity matrix can transform the eigenvalue factorization in spectral clustering with computational complexity O(n3) into the singular value decomposition (SVD) with O(nc2) time cost, where n and c , respectively, denote the numbers of samples and classes. Thus, the computational efficiency can be improved. Moreover, in order to learn a better clustering model, we set that the constructed similarity graph approximates each view affinity graph as close as possible by adding the constraint as the initial affinity matrices own. Furthermore, substantial experiments are conducted, which verifies the superiority of the proposed two clustering methods comparing with single-view clustering approaches and state-of-the-art multi-view clustering methods.

Multiple-View Spectral Embedded Clustering Using a Co-training Approach

Multi-View Clustering Via Simultaneously Learning Shared Subspace And Affinity Matrix

Adaptive Graph Fusion Learning for Multi-View Spectral Clustering

Multi-view Clustering Via Spectral Embedding Fusion

One-step Multi-View Spectral Clustering by Learning Common and Specific Nonnegative Embeddings

Joint Learning of Latent Similarity and Local Embedding for Multi-View Clustering

Multiple Kernel Learning Based Multi-view Spectral Clustering

Multi-view Clustering by Joint Spectral Embedding and Spectral Rotation

Multi-view Spectral Clustering Via Integrating Nonnegative Embedding and Spectral Embedding

Lifelong Multi-view Spectral Clustering

Multiple non-redundant spectral clustering views

One for all: A novel Dual-space Co-training baseline for Large-scale Multi-View Clustering

Multiview Data Clustering by using Adaptive Spectral Co-clustering

Multi-View Clustering Via Nonnegative and Orthogonal Graph Reconstruction.

Adaptive spectral co-clustering for multiview data

Multi-view co-clustering with multi-similarity

Unified Low-Rank Tensor Learning and Spectral Embedding for Multi-View Subspace Clustering

Multiple Partitions Alignment Via Spectral Rotation

One-step Multi-View Spectral Clustering Based on Multi-Feature Similarity Fusion

Incomplete Multiview Spectral Clustering With Adaptive Graph Learning

Unified Multi-View Orthonormal Non-Negative Graph Based Clustering Framework