Search papers, labs, and topics across Lattice.
This paper introduces Perron--Frobenius Operator Matching (PFOM), a novel generative modeling framework that utilizes the integral PF operator to match density evolution across various model types, including flow, diffusion, and jump models. The authors demonstrate that Kullback--Leibler divergence uniquely maintains equality between density-level and sample-conditioned objectives, leading to a practical loss function that aligns with Koopman path matching. Empirical results on Gaussian mixtures and two-moons show that PFOM outperforms existing methods in terms of KL, Wasserstein, and MMD metrics while enhancing computational efficiency through Nesterov-accelerated training and sampling techniques.
Kullback鈥揕eibler divergence is the secret sauce that ensures density-level and sample-conditioned objectives align perfectly in generative modeling.
We introduce Perron--Frobenius Operator Matching (PFOM), a generative framework that matches density evolution via the integral PF operator, subsuming flow, diffusion, and jump models. We prove that among Bregman divergences, only Kullback--Leibler divergence preserves equality between density-level and sample-conditioned objectives, yielding a practical loss equivalent to Koopman path matching. We further develop Nesterov-accelerated training and sampling that stabilize discretization and accelerate convergence. %On Gaussian mixtures and two-moons, PFOM achieves faster KL/$W_2$/MMD decrease and improved wall-clock efficiency with empirical validation. PFOM unifies operator-theoretic identification with modern generative modeling and opens paths to adaptive dictionaries and high-dimensional applications.
