A Riemannian gossip approach to decentralized matrix completion
This addresses scalable and parallelizable matrix completion for distributed systems, but appears incremental as it builds on existing gossip and manifold methods.
The paper tackles decentralized matrix completion by proposing gossip algorithms on the Riemannian Grassmann manifold, enabling local processing by agents while achieving consensus on global low-rank factors, with numerical experiments demonstrating good performance on benchmarks.
In this paper, we propose novel gossip algorithms for the low-rank decentralized matrix completion problem. The proposed approach is on the Riemannian Grassmann manifold that allows local matrix completion by different agents while achieving asymptotic consensus on the global low-rank factors. The resulting approach is scalable and parallelizable. Our numerical experiments show the good performance of the proposed algorithms on various benchmarks.