Track manipulation, locomotion and embodied systems.
Search papers, labs, and topics across Lattice.
This paper introduces generalized mass-preserving coupled flow matching (G-MPCF) to address the issue of inaccurate regional probability distributions in generative flows, which can distort population statistics and compromise inferences about rare events. G-MPCF achieves this by matching probability masses between Gaussian source regions and weighted data regions, directly encoding desired regional frequencies into the learned dynamics. Experimental results demonstrate significant reductions in relative score error and numerical integration error on a rare-mode benchmark, and improved class-mass error and image fidelity on CIFAR-10 compared to existing methods.
Generative flows can now precisely capture rare event frequencies, slashing relative score error by 95% and class-mass error by 44.6% by directly matching regional probability masses.
Generative models offer a powerful route to scientific simulation and discovery by enabling the study of complex populations and rare phenomena. Flow matching can generate realistic samples, yet inaccurate occurrence frequencies can distort population statistics and compromise inferences about rare events. Here we developed generalized mass-preserving coupled flow matching (G-MPCF), which matches the probability masses of Gaussian source regions to those of weighted data regions. This construction encodes the desired regional frequencies in the correspondence between noise and observations, shaping the learned dynamics while preserving the prescribed target distribution. On a rare-mode benchmark, G-MPCF reduces relative score error by 95.0% and numerical integration error by 89.9% compared with independent flow matching. On CIFAR-10 with a 100:1 class-probability ratio, it reduces corrected class-mass error by 44.6% compared with minibatch optimal transport, while improving mean image fidelity with matched computational budgets. We envision that incorporating regional probability calibration into generative modelling will support more reliable scientific simulation and inference, enabling the study of systems in which rare events and their occurrence frequencies are central to understanding the underlying mechanisms.