Incomplete Multi-View Clustering via Neighborhood-Conditioned Diffusion

Incomplete multi-view clustering (IMVC) aims to uncover shared clustering structures from heterogeneous views with partial observations. Recently, existing generative IMVC methods have made significant progress in this field; however, they still remain limited in two aspects. On the one hand, they rely on weak cross-view signals, resulting in unstable latent recovery when facing missing data. On the other hand, they overlook stable cross-view neighborhood structures, leading to weak structural constraint. To address these limitations, we propose neighborhood-conditioned diffusion for incomplete multi-view clustering (IMVC-NCD), which achieves robust latent completion. Our method learns compact view-specific latent representations and constructs a unified conditioning vector by aggregating stable local neighborhood structures from available views while encoding missingness states, providing reliable guidance for diffusion-based denoising. With neighborhood-level conditioning, IMVC-NCD produces semantically aligned and view-consistent latent representations that are well suited for clustering, even under high missing-view ratios. Extensive experiments on four benchmark datasets demonstrate the effectiveness and robustness of our method compared with state-of-the-art IMVC approaches. Our code is available at https://github.com/zgh1115/IMVC-NCD.

What can we do when different descriptions of the same data are incomplete? For example, an object may have an image, text, or other features, but some of them may be missing. We study how to make clustering reliable in this common setting, where the goal is to group similar samples without labels. Existing methods try to recover the missing information, but they often lack strong guidance from the available views. They may also miss an important clue: nearby samples can still reveal stable relationships about the underlying data structure. Our paper proposes IMVC-NCD, a diffusion-based method that uses these neighborhood relationships to guide missing-information recovery. It combines stable local neighborhoods from observed views with information about which views are missing, helping produce more consistent representations for clustering. Our results show that this strategy improves clustering accuracy and robustness, especially when many views are missing.