Distributed Multi-View Subspace Clustering via Auto-Weighted Spectral Embedding

This work examines a distributed multi-view clustering problem, where the data associated with different views is stored across multiple edge devices. A sparse subspace clustering method is adopted using auto-weighted spectral embedding to ensure that the clustering solution is consistent among local edge devices. A master-slave architecture is adopted where clustering is first performed separately at the edge devices based on their local single-view datasets but are coordinated by a spectral regularizer computed at the central node. The optimization is performed using an alternating optimization approach, where the local self-representation and the global cluster indicator matrices are optimized in turn until convergence. The weighting of the regularizer is updated in each iteration of the process and adapts automatically to the fit of the spectral embedding at different locations. The proof of convergence is provided, followed by experimental results on two public datasets, namely, Extended Yale-B and IXMAS, which demonstrate the effectiveness of the proposed method.