2025/01/10 by Julius Berner, Berner, Julius, Lorenz Richter +7 · 5 citations
Computer Science · Mathematics · #Advanced Mathematical Modeling in Engineering #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Mathematical Biology Tumor Growth
paper · pdf · doi:10.48550/arxiv.2501.06148
openalex publication_date 2025/01/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We study the problem of training neural stochastic differential equations, or diffusion models, to sample from a Boltzmann distribution without access to target samples. Existing methods for training such models enforce time-reversal of the generative and noising processes, using either differentiable simulation or off-policy reinforcement learning (RL). We prove equivalences between families of objectives in the limit of infinitesimal discretization steps, linking entropic RL methods (GFlowNets) with continuous-time objects (partial differential equations and path space measures). We further show that an appropriate choice of coarse time discretization during training allows greatly improved sample efficiency and the use of time-local objectives, achieving competitive performance on standard sampling benchmarks with reduced computational cost.