← Back to all news
arXiv Machine Learning By Ewsey Obzherin, Valentin Khrulkov

Density-Level Backward Error for Low-NFE Generative Flows

Read the original on arXiv Machine Learning →

The paper introduces a density‑level backward‑error framework for low‑NFE generative ODE sampling, identifying the part of a solver’s modified‑equation defect that alters the evolving marginal while ignoring solenoidal trajectory distortion. This leads to a solver‑order‑aware compute‑allocation rule and an objective for choosing numerically favorable velocity representatives, distinct from schedule search or higher‑order solver design. Experiments on Gaussian, mixture models, and a pretrained CIFAR‑10 EDM probability‑flow ODE show that density‑aware allocation consistently improves fixed‑update Euler sampling in the low‑NFE regime and accelerates convergence to a dense Euler reference.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
Jul 17

Trajectory-Aware Flow Matching for Topology Optimisation

arXiv:2607. 14652v1 Announce Type: new Abstract: Topology optimisation (TO) often requires repeated finite element analysis and sensitivity-based material updates, which can be costly when multiple candidate designs are needed under varying physical and design conditions.

By Shusheng Xiao, Jinshuai Bai, Hyogu Jeong, Yunfei Xi, Yilin Gui, YuanTong Gu
diffusionsafety
More like this →
arXiv Machine Learning
Sep 15

A Variational Optimal Transport Operator on Incompressible Flow

The paper introduces the Variational Incompressible Optimal Transport (VIOT) operator, a generative neural operator that predicts divergence‑free velocity fields for incompressible density transport. VIOT combines a stream‑function representation, a regularized transport objective, and a Fourier Neural Operator backbone to amortize the solve across new source‑target pairs and grid resolutions. Experiments on 2D and 3D benchmarks show that VIOT produces full transport trajectories in seconds, achieving roughly a $10^4 imes$ speedup over per‑instance baselines that require hours of optimization.

By Jinjin He, Shenyifan Lu, Sinan Wang, Zhiqi Li, Duowen Chen, Bo Zhu
benchmarks
More like this →
arXiv Machine Learning
Aug 10

Limit Points of Reflow with Minibatch Optimal Transport

arXiv:2608. 07042v1 Announce Type: cross Abstract: Rectified flows, also called flow matching or stochastic interpolants, are generative models that learn a time-dependent vector field steering a probability curve between two probability distributions, usually referred to as latent and target distributions.

By Antonin Chambolle, Johannes Hertrich
diffusion
More like this →
arXiv AI
Sep 2

Training-Free Refinement of Flow Matching with Divergence-based Sampling

The paper introduces Flow Divergence Sampler (FDS), a training‑free method that refines intermediate states in flow‑matching models by using the divergence of the marginal velocity field to detect and correct misguidance toward low‑density regions. FDS operates during inference, requires no additional training, and can be applied as a plug‑and‑play module with standard solvers and existing flow backbones. Experiments show that FDS consistently improves fidelity in tasks such as text‑to‑image synthesis and inverse problems.

By Yeonwoo Cha, Jaehoon Yoo, Semin Kim, Yunseo Park, Jinhyeon Kwon, Seunghoon Hong
diffusion
More like this →