Discretely-Refined Multi-view Clustering via Aligned Anchor Learning

Anchor-based multi-view clustering has garnered wide attention for its ability to reduce the computational complexity of large-scale spectral clustering. However, existing methods mostly adopt a unidirectional optimization paradigm confined to sample-anchor bipartite graphs, treating the construction of the consensus graph and discrete clustering assignments as separate sub-problems to be solved independently. This weakens the information exchange between continuous representation and discrete structure, confining the optimization process to iterative updates within local modules.To address these limitations, we propose a Discretely-Refined Multi-view Clustering(DRMC) via Aligned Anchor Learning. Unlike approaches that directly perform fusion in the anchor space, our method starts from the anchor graph, elevates sample-anchor associations to sample-level similarity graph representations, and thereby enhances both within-cluster similarity and between-cluster separation. Furthermore, we design a discrete feedback module that jointly conducts spectral embedding learning and discrete label assignment by orthogonally aligning the continuous embedding matrix with the discrete indicator matrix. The resulting discrete partition is then fed back into the consensus graph construction, continuously refining the graph structure. Experiments on multiple benchmark datasets demonstrate that the proposed method exhibits significant advantages over existing state-of-the-art approaches.

Real-world data often comes from several different sources or perspectives. For example, the same news event may include text, images, and audio, and each source can provide useful but different information. Multi-view clustering aims to automatically group such data without human-provided labels. This is useful for tasks such as organizing multimedia data, analyzing medical images, and finding hidden patterns in large datasets. Existing efficient clustering methods often use a small number of representative points to reduce computation. However, many of them first build a data relationship graph and then generate the final groups in a separate step. Because these two steps do not fully communicate with each other, the learned graph may contain noisy, which can reduce clustering accuracy. This paper proposes DRMC, a new method that allows the learned groups to help improve the data relationship graph during training. Instead of only relying on rough relationships between data and representative points, DRMC builds a more detailed sample-level graph and then refines it using the current grouping results. Experiments on multiple benchmark datasets show that DRMC can produce more reliable clustering results than many existing methods. The results suggest that using clustering results as feedback is an effective way to improve multi-view clustering.