Multisample Flow Matching: Straightening Flows with Minibatch Couplings

04/28/2023
by   Aram-Alexandre Pooladian, et al.
1

Simulation-free methods for training continuous-time generative models construct probability paths that go between noise distributions and individual data samples. Recent works, such as Flow Matching, derived paths that are optimal for each data sample. However, these algorithms rely on independent data and noise samples, and do not exploit underlying structure in the data distribution for constructing probability paths. We propose Multisample Flow Matching, a more general framework that uses non-trivial couplings between data and noise samples while satisfying the correct marginal constraints. At very small overhead costs, this generalization allows us to (i) reduce gradient variance during training, (ii) obtain straighter flows for the learned vector field, which allows us to generate high-quality samples using fewer function evaluations, and (iii) obtain transport maps with lower cost in high dimensions, which has applications beyond generative modeling. Importantly, we do so in a completely simulation-free manner with a simple minimization objective. We show that our proposed methods improve sample consistency on downsampled ImageNet data sets, and lead to better low-cost sample generation.

READ FULL TEXT

page 1

page 6

page 9

page 15

page 18

page 19

page 20

research
10/06/2022

Flow Matching for Generative Modeling

We introduce a new paradigm for generative modeling built on Continuous ...
research
09/07/2022

Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow

We present rectified flow, a surprisingly simple approach to learning (n...
research
06/11/2023

On Kinetic Optimal Probability Paths for Generative Models

Recent successful generative models are trained by fitting a neural netw...
research
06/26/2023

Equivariant flow matching

Normalizing flows are a class of deep generative models that are especia...
research
07/07/2023

Simulation-free Schrödinger bridges via score and flow matching

We present simulation-free score and flow matching ([SF]^2M), a simulati...
research
05/11/2023

Towards Understanding and Improving GFlowNet Training

Generative flow networks (GFlowNets) are a family of algorithms that lea...
research
07/11/2022

Matching Normalizing Flows and Probability Paths on Manifolds

Continuous Normalizing Flows (CNFs) are a class of generative models tha...

Please sign up or login with your details

Forgot password? Click here to reset