Neural Flow Samplers with Shortcut Models
This addresses sampling problems in applications like posterior inference and molecular dynamics, but appears incremental as it builds on existing flow-based neural samplers.
The paper tackles the challenge of sampling from unnormalized densities by introducing an improved estimator for intractable terms and a shortcut consistency model to reduce sampling steps. The proposed Neural Flow Shortcut Sampler outperforms existing flow-based neural samplers on synthetic datasets and complex n-body system targets.
Sampling from unnormalized densities presents a fundamental challenge with wide-ranging applications, from posterior inference to molecular dynamics simulations. Continuous flow-based neural samplers offer a promising approach, learning a velocity field that satisfies key principles of marginal density evolution (e.g., the continuity equation) to generate samples. However, this learning procedure requires accurate estimation of intractable terms linked to the computationally challenging partition function, for which existing estimators often suffer from high variance or low accuracy. To overcome this, we introduce an improved estimator for these challenging quantities, employing a velocity-driven Sequential Monte Carlo method enhanced with control variates. Furthermore, we introduce a shortcut consistency model to boost the runtime efficiency of the flow-based neural sampler by minimizing its required sampling steps. Our proposed Neural Flow Shortcut Sampler empirically outperforms existing flow-based neural samplers on both synthetic datasets and complex n-body system targets.