首页> 外文会议>Conference on uncertainty in artificial intelligence >Max-Product Belief Propagation for Linear Programming: Applications to Combinatorial Optimization
【24h】

Max-Product Belief Propagation for Linear Programming: Applications to Combinatorial Optimization

机译:线性编程的最大产品信念传播:组合优化的应用

获取原文

摘要

Max-product belief propagation (BP) is a popular message-passing algorithm for computing a maximum-a-posteriori (MAP) assignment in a joint distribution represented by a graphical model (GM). It has been shown that BP can solve a few classes of Linear Programming (LP) formulations to combinatorial optimization problems including maximum weight matching and shortest path, i.e., BP can be a distributed solver for certain LPs. However, those LPs and corresponding BP analysis are very sensitive to underlying problem setups, and it has been not clear what extent these results can be generalized to. In this paper, we obtain a generic criteria that BP converges to the optimal solution of given LP, and show that it is satisfied in LP formulations associated to many classical combinatorial optimization problems including maximum weight perfect matching, shortest path, traveling salesman, cycle packing and vertex cover. More importantly, our criteria can guide the BP design to compute fractional LP solutions, while most prior results focus on integral ones. Our results provide new tools on BP analysis and new directions on efficient solvers for large-scale LPs.
机译:Max-Product信念传播(BP)是一种流行的消息传递算法,用于计算由图形模型(GM)表示的联合分布中的最大-A-Bouthiori(MAP)分配。已经表明,BP可以解决几类线性编程(LP)制剂,包括组合优化问题,包括最大权重匹配和最短路径,即BP可以是某些LPS的分布式求解器。但是,这些LPS和相应的BP分析对潜在的问题设置非常敏感,并且目前尚不清楚这些结果可以推广到什么程度。在本文中,我们获得了BP收敛于给定LP的最佳解决方案的通用标准,并表明它在与许多经典组合优化问题相关的LP配方中满足,包括最大重量完美匹配,最短路径,旅行推销员,循环包装和顶点盖子。更重要的是,我们的标准可以指导BP设计来计算分数LP解决方案,而大多数事先结果侧重于整体。我们的结果为大型LPS提供了关于BP分析和高效求解器的新方向的新工具。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号