0


通过双向多重切割的二进制MRF中的MAP估计

MAP estimation in Binary MRFs via Bipartite Multi-cuts
课程网址: http://videolectures.net/nips2010_reddi_map/  
主讲教师: Sashank Jakkam Reddi
开课单位: 印度理工学院
开课时间: 2011-01-13
课程语种: 英语
中文简介:
我们提出了一种新的LP弛豫方法来获得具有成对势的二元MRF的映射分配。我们的松弛是从将映射分配问题简化为最近提出的二部多割问题的一个实例得到的,其中LP松弛保证提供一个对数k近似,其中k是MRF中邻近非子模边的顶点数。然后,我们提出了一种组合算法来有效地求解LP,并通过在epsilon近似下同时求解其对偶到来提供一个下界。该算法比最新的消息传递算法快一个数量级,提供了更好的地图分数和边界,该算法使用三阶边界约束来收紧局部边缘多面体。
课程简介: We propose a new LP relaxation for obtaining the MAP assignment of a binary MRF with pairwise potentials. Our relaxation is derived from reducing the MAP assignment problem to an instance of a recently proposed Bipartite Multi-cut problem where the LP relaxation is guaranteed to provide an log k approximation where k is the number of vertices adjacent to non-submodular edges in the MRF. We then propose a combinatorial algorithm to efficiently solve the LP and also provide a lower bound by concurrently solving its dual to within an epsilon approximation. The algorithm is up to an order of magnitude faster and provides better MAP scores and bounds than the state of the art message passing algorithm that tightens the local marginal polytope with third-order marginal constraints.
关 键 词: 计算机科学; 算法和数据结构; 数学; 图论
课程来源: 视频讲座网
最后编审: 2020-06-15:heyf
阅读次数: 27