arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1902.00282stat.MLcs.LG
收藏

将MCMC动力学理解为Wasserstein空间上的流

Understanding MCMC Dynamics as Flows on the Wasserstein Space

Tsinghua University(清华大学)

Chang Liu, Jingwei Zhuo, Jun Zhu

摘要:

MCMC中使用的Langevin动力学是Wasserstein空间上库尔贝克-莱布勒散度的梯度流,这有助于收敛性分析并启发了最近的基于粒子的变分推断方法(ParVI).但没有更多的MCMC动力学是以这种方式理解的。在这项工作中,通过开发新的概念,我们提出了一个理论框架,承认一个一般的MCMC动力学的纤维梯度哈密顿流的Wasserstein空间的纤维黎曼泊松流形。流的“守恒+收敛”结构给出了一般MCMC动力学行为的清晰图像。该框架还使ParVI模拟MCMC动态,这丰富了ParVI家族更有效的动态,并适应ParVI优势MCMC。我们开发了两个ParVI方法,一个特定的MCMC动态,并在实验中证明的好处。

英文摘要:

It is known that the Langevin dynamics used in MCMC is the gradient flow of the KL divergence on the Wasserstein space, which helps convergence analysis and inspires recent particle-based variational inference methods (ParVIs). But no more MCMC dynamics is understood in this way. In this work, by developing novel concepts, we propose a theoretical framework that recognizes a general MCMC dynamics as the fiber-gradient Hamiltonian flow on the Wasserstein space of a fiber-Riemannian Poisson manifold. The "conservation + convergence" structure of the flow gives a clear picture on the behavior of general MCMC dynamics. The framework also enables ParVI simulation of MCMC dynamics, which enriches the ParVI family with more efficient dynamics, and also adapts ParVI advantages to MCMCs. We develop two ParVI methods for a particular MCMC dynamics and demonstrate the benefits in experiments.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑