analiza aiqbench
This paper introduces UW-Flow, an improvement to teacher-free generative models that addresses marginal constraint issues in finite-batch training through unbalanced optimal transport. By replacing balanced Sinkhorn divergence with a debiased variant, the method reduces velocity error and enhances generation quality on standard benchmarks like MNIST and ImageNet. The work offers theoretical insights into how relaxing exact marginals can better handle minibatch fluctuations without sacrificing model identifiability.
Unbalanced W-Flow: Unbalanced Wasserstein Drifting under Finite-Batch Marginal Fluctuations
arXiv:2610.18552v1 Announce Type: new Abstract: Teacher-free one-step generative models such as W-Flow learn directly from distributional objectives, with the Sinkhorn divergence providing the strongest empirical performance among the energies studied in W-Flow. However, Sinkhorn divergence is based on balanced optimal transport, whose exact marginal constraints can induce undesirable transport under finite minibatch fluctuations. We introduce UW-Flow, which replaces the balanced Sinkhorn energy with a debiased unbalanced Sinkhorn divergence and derives the corresponding Wasserstein velocity. The resulting dynamics retain W-Flow's cross-minus-self structure, recover the balanced limit, and preserve target identifiability under suitable assumptions. We further characterize the finite-batch velocity error of balanced and unbalanced transport by comparing each minibatch field with its population counterpart. Our analysis identifies conditions under which marginal relaxation yields smaller minibatch-to-population velocity error, and shows that for separated matched modes, balanced sensitivity grows with inter-mode separation while the corresponding Unbalanced Optimal Transport (UOT) sensitivity remains controlled by local transport geometry. Controlled experiments on fixed-support and continuous two-mode distributions support these predictions. On MNIST, CIFAR-10, and ImageNet-100, UW-Flow improves generation under matched training budgets, with the largest gains in unconditional generation and consistent improvements in the class-conditional setting.
Abstract:Teacher-free one-step generative models such as W-Flow learn directly from distributional objectives, with the Sinkhorn divergence providing the strongest empirical performance among the energies studied in W-Flow. However, Sinkhorn divergence is based on balanced optimal transport, whose exact marginal constraints can induce undesirable transport under finite minibatch fluctuations. We introduce UW-Flow, which replaces the balanced Sinkhorn energy with a debiased unbalanced Sinkhorn divergence and derives the corresponding Wasserstein velocity. The resulting dynamics retain W-Flow's cross-minus-self structure, recover the balanced limit, and preserve target identifiability under suitable assumptions. We further characterize the finite-batch velocity error of balanced and unbalanced transport by comparing each minibatch field with its population counterpart. Our analysis identifies conditions under which marginal relaxation yields smaller minibatch-to-population velocity error, and shows that for separated matched modes, balanced sensitivity grows with inter-mode separation while the corresponding Unbalanced Optimal Transport (UOT) sensitivity remains controlled by local transport geometry. Controlled experiments on fixed-support and continuous two-mode distributions support these predictions. On MNIST, CIFAR-10, and ImageNet-100, UW-Flow improves generation under matched training budgets, with the largest gains in unconditional generation and consistent improvements in the class-conditional setting.
Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2610.18552 [cs.LG]
(or arXiv:2610.18552v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.18552 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 285 cuvinte
Subiecte conexe
One-Step Generative Modeling via Unbalanced Optimal Transport
arXiv · 30 sept. 2026
Discrete Wasserstein Flows for One-Step Generative Modeling
arXiv · 2 oct. 2026
Inference meta-monitoring for Amazon SageMaker AI endpoints with Amazon Quick
AWS · 30 iul. 2026
Friction-Augmented Drifting Models for Resource-Efficient Domain Translation
arXiv · 17 aug. 2026