Improving Rectified Flow with Boundary Conditions
This addresses a specific limitation in generative modeling for image generation, though it appears incremental as it modifies an existing method with minimal code changes.
The paper tackles the problem of inaccurate velocity field estimation in Rectified Flow generative models by proposing a Boundary-enforced Rectified Flow Model that enforces boundary conditions, resulting in an 8.01% improvement in FID score on ImageNet with ODE sampling and 8.98% improvement with SDE sampling.
Rectified Flow offers a simple and effective approach to high-quality generative modeling by learning a velocity field. However, we identify a limitation in directly modeling the velocity with an unconstrained neural network: the learned velocity often fails to satisfy certain boundary conditions, leading to inaccurate velocity field estimations that deviate from the desired ODE. This issue is particularly critical during stochastic sampling at inference, as the score function's errors are amplified near the boundary. To mitigate this, we propose a Boundary-enforced Rectified Flow Model (Boundary RF Model), in which we enforce boundary conditions with a minimal code modification. Boundary RF Model improves performance over vanilla RF model, demonstrating 8.01% improvement in FID score on ImageNet using ODE sampling and 8.98% improvement using SDE sampling.