Low-Pass Flow Matching
2026-06-01 • Machine Learning
Machine Learning
AI summaryⓘ
The authors noticed that typical flow matching methods use white noise, which doesn't match the usual frequency patterns of real data that lose strength at higher frequencies. They created Low-Pass Flow Matching, which gradually changes the noise to better fit these real-world patterns. Their tests on image generation, including galaxy images, show that this method works well, especially when used with certain ODE solvers, giving similar or better quality samples while being faster. This means their approach better matches the data's natural characteristics and reduces computation time.
Flow MatchingWhite NoiseLow-Pass FilterFrequency SpectrumImage GenerationODE SolversSpectral BiasAdaptive SamplingGalaxy10 Dataset
Authors
Francesco M. Ruscio, T. Konstantin Rusch
Abstract
Flow Matching typically relies on white noise sources, a choice often misaligned with the power spectra of natural data, which tend to decay with frequency. To address this, we introduce Low-Pass Flow Matching, a variant of Flow Matching based on an operator-modulated interpolant. This formulation induces a time-varying spectral bias that transitions from the source spectrum to a frequency-decaying bias as the path approaches the data. We validate our method on unconditional image generation tasks, including the scientific Galaxy10 dataset. Empirically, we show that our method is particularly effective when paired with adaptive ODE solvers, where it improves or preserves sample quality while substantially reducing sampling cost compared to standard baselines.