Straight-Path Flow Matching for Incomplete Multi-View Clustering

2026-07-07

AI Summary

This work addresses the challenge of clustering with incomplete multi-view data by proposing an end-to-end framework based on straight-path flow matching. The method completes missing views by constructing a deterministic ordinary differential equation (ODE) flow between observed and unobserved views, and enhances cross-view clustering consistency through cluster-level alignment and entropy regularization. Notably, it is the first to apply straight probability paths to incomplete multi-view clustering and provides theoretical justification that, compared to diffusion models, deterministic flows better preserve cluster structure under finite-step integration—making them more aligned with clustering objectives. The approach achieves state-of-the-art performance on standard incomplete multi-view clustering (IMVC) benchmarks.

Abstract

Incomplete Multi-View Clustering addresses the problem of clustering multi-modal data when certain views are missing. Recent end-to-end generative approaches leverage diffusion models to recover missing views via stochastic noise-to-data trajectories. While expressive, such mechanisms are not explicitly designed for clustering, as they initialize from cluster-agnostic noise and rely on stochastic denoising dynamics. In this work, we revisit probability path design in end-to-end generative IMVC. We introduce a flow-matching framework with a linear interpolation path between paired view representations, that replaces diffusion with probability flows between observed and missing views. We provide a formal analysis showing that deterministic ODE flows are inherently better aligned with clustering objectives than diffusion-based stochastic trajectories, especially in terms of transport mechanisms that respect class-conditional data distributions and maintain cluster consistency in finite-step regimes. Building upon this insight, we develop an end-to-end IMVC architecture that integrates straight-path flow-matching view completion with cluster-level and entropy-based alignment to enforce cross-view clustering consistency. Extensive experiments on standard IMVC benchmarks demonstrate that the proposed framework establishes new state-of-the-art performance.

Problem

• Incomplete Multi-View Clustering
• Clustering
• Missing Views
• Generative Models
• Diffusion Models

Innovation

• flow matching
• incomplete multi-view clustering
• deterministic ODE
• straight-path interpolation
• cross-view consistency