Author:
Lehmann Tobias,von Renesse Max-K.,Sambale Alexander,Uschmajew André
Abstract
AbstractWe derive an a priori parameter range for overrelaxation of the Sinkhorn algorithm, which guarantees global convergence and a strictly faster asymptotic local convergence. Guided by the spectral analysis of the linearized problem we pursue a zero cost procedure to choose a near optimal relaxation parameter.
Funder
Max Planck Institute for Mathematics in the Sciences
Publisher
Springer Science and Business Media LLC
Reference19 articles.
1. Cuturi, M.: Sinkhorn distances: lightspeed computation of optimal transport. In: Burges, C.J.C., et al. (eds.) Advances in Neural Information Processing Systems, vol. 26. Curran Associates, Inc., pp. 2292–2300 (2013)
2. Sinkhorn, R.: Diagonal equivalence to matrices with prescribed row and column sums. Am. Math. Monthly 74, 402–405 (1967)
3. Franklin, J., Lorenz, J.: On the scaling of multidimensional matrices. Linear Algebra Appl. 114(115), 717–735 (1989)
4. Peyré, G., Cuturi, M.: Computational optimal transport. Found. Trends Mach. Learn. 11(5–6), 355–607 (2019)
5. Thibault, A., Chizat, L., Dossal, C., Papadakis, N.: Overrelaxed Sinkhorn–Knopp algorithm for regularized optimal transport. Algorithms 14(5), 143 (2021)
Cited by
2 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献