We tackle the problem of sampling from intractable high-dimensional density functions, a fundamental task that often appears in machine learning and statistics. We extend recent sampling-based approaches that leverage controlled stochastic processes to model approximate samples from these target densities. The main drawback of these approaches is that the training objective requires full trajectories to compute, resulting in sluggish credit assignment issues due to use of entire trajectories and a learning signal present only at the terminal time. In this work, we present Diffusion Generative Flow Samplers (DGFS), a sampling-based framework where the learning process can be tractably broken down into short partial trajectory segments, via parameterizing an additional "flow function". Our method takes inspiration from the theory developed for generative flow networks (GFlowNets), allowing us to make use of intermediate learning signals and benefit from off-policy exploration capabilities. Through a variety of challenging experiments, we demonstrate that DGFS results in more accurate estimates of the normalization constant than closely-related prior methods.

翻译：我们解决从难以处理的高维密度函数中采样的问题，这是机器学习和统计学中常见的基础任务。我们扩展了近期基于采样的方法，这些方法利用受控随机过程来模拟这些目标密度的近似样本。这些方法的主要缺点是训练目标需要计算完整轨迹，导致由于使用整个轨迹且仅在终端时刻存在学习信号而出现迟缓的信用分配问题。在这项工作中，我们提出了扩散生成流采样器（DGFS），这是一种基于采样的框架，通过参数化额外的“流函数”，学习过程可被可处理地分解为短的部分轨迹片段。我们的方法借鉴了为生成流网络（GFlowNets）发展的理论，使我们能够利用中间学习信号并受益于离策略探索能力。通过一系列具有挑战性的实验，我们证明DGFS相比密切相关的先前方法能更准确地估计归一化常数。