Cutting Plane Algorithms for Variational Inference in Graphical Models
msra(2007)
摘要
Abstract In this thesis, we give a new class of outer bounds on the marginal polytope, and propose a cutting-plane algorithm for eciently,optimizing over these constraints. When combined with a concave upper bound on the entropy, this gives a new vari- ational inference algorithm for probabilistic inference in discrete Markov Random
更多查看译文
关键词
graphical model,upper bound
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要