BriefGPT.xyz
Sep, 2022
稀疏约束的最优输运
Sparsity-Constrained Optimal Transport
HTML
PDF
Tianlin Liu, Joan Puigcerver, Mathieu Blondel
TL;DR
本文提出了一种新的最优输运问题方法,具有明确的基数约束,既能保证传送方案的稀疏性,又能够限制匹配的最大数量,使其在稀疏专家混合等应用中能够更好地解决计算性能问题。
Abstract
regularized optimal transport
(OT) is now increasingly used as a loss or as a matching layer in neural networks.
entropy-regularized ot
can be computed using the Sinkhorn algorithm but it leads to fully-dense tra
→