Conservative Flows: A New Paradigm of Generative Models
This work offers a new paradigm for generative models that can leverage existing flow models, potentially improving generation quality for image synthesis tasks.
The authors propose a new generative modeling paradigm where discrete stochastic dynamics preserve the data distribution, initialized from data-supported states rather than noise. Their samplers improve over original generation procedures on ImageNet-256 and Oxford Flowers-102.
Modern generative modeling is dominated by transport from a noise prior to data. We propose an alternative paradigm in which generation is performed by a discrete stochastic dynamics that leaves the data distribution invariant, initialized from data-supported states rather than from noise. The framework can utilize any pretrained flow model. We develop two probability-preserving sampling mechanisms, a corrected Langevin dynamics with a Metropolis adjustment and a predictor-corrector flow, that operate directly on existing checkpoints. We validate the framework on a synthetic Swiss-roll target, ImageNet-256 and Oxford Flowers-102, where our samplers consistently improve over the original generation procedures.