On Scalable and Efficient Computation of Large Scale Optimal Transport
This work addresses the scalability issue for researchers and practitioners using OT in applications like domain adaptation, though it is incremental as it builds on existing generative and optimization methods.
The authors tackled the computational scalability problem of Optimal Transport (OT) in machine learning by proposing SPOT, an implicit generative learning-based framework that approximates the transport plan and solves OT efficiently using primal-dual stochastic gradient algorithms, showing robust performance and favorable convergence in experiments.
Optimal Transport (OT) naturally arises in many machine learning applications, yet the heavy computational burden limits its wide-spread uses. To address the scalability issue, we propose an implicit generative learning-based framework called SPOT (Scalable Push-forward of Optimal Transport). Specifically, we approximate the optimal transport plan by a pushforward of a reference distribution, and cast the optimal transport problem into a minimax problem. We then can solve OT problems efficiently using primal dual stochastic gradient-type algorithms. We also show that we can recover the density of the optimal transport plan using neural ordinary differential equations. Numerical experiments on both synthetic and real datasets illustrate that SPOT is robust and has favorable convergence behavior. SPOT also allows us to efficiently sample from the optimal transport plan, which benefits downstream applications such as domain adaptation.