Hierarchical Refinement: Optimal Transport to Infinity and Beyond
Halmos, Peter, Gold, Julian, Liu, Xinhao, Raphael, Benjamin J.
–arXiv.org Artificial Intelligence
Optimal transport (OT) has enjoyed great success in machine-learning as a principled way to align datasets via a least-cost correspondence. This success was driven in large part by the runtime efficiency of the Sinkhorn algorithm [Cuturi 2013], which computes a coupling between points from two datasets. However, Sinkhorn has quadratic space complexity in the number of points, limiting the scalability to larger datasets. Low-rank OT achieves linear-space complexity, but by definition, cannot compute a one-to-one correspondence between points. When the optimal transport problem is an assignment problem between datasets then the optimal mapping, known as the Monge map, is guaranteed to be a bijection. In this setting, we show that the factors of an optimal low-rank coupling co-cluster each point with its image under the Monge map. We leverage this invariant to derive an algorithm, Hierarchical Refinement (HiRef), that dynamically constructs a multiscale partition of a dataset using low-rank OT subproblems, culminating in a bijective coupling. Hierarchical Refinement uses linear space and has log-linear runtime, retaining the space advantage of low-rank OT while overcoming its limited resolution. We demonstrate the advantages of Hierarchical Refinement on several datasets, including ones containing over a million points, scaling full-rank OT to problems previously beyond Sinkhorn's reach.
arXiv.org Artificial Intelligence
Mar-4-2025
- Country:
- Asia > Russia (0.04)
- Europe
- Russia (0.04)
- Middle East > Malta
- Northern Region > Northern District > Mosta (0.04)
- France > Grand Est
- Meurthe-et-Moselle > Nancy (0.04)
- Genre:
- Research Report (0.64)
- Industry:
- Health & Medicine (0.68)
- Technology: