0


带组合约束的在线子模极小化

Online submodular minimization with combinatorial constraints
课程网址: http://videolectures.net/nipsworkshops2010_jegelka_osm/  
主讲教师: Stefanie Jegelka
开课单位: 马克斯普朗克研究所
开课时间: 2011-01-13
课程语种: 英语
中文简介:

最近,人们越来越感兴趣的是用次模块成本函数代替组合问题中的线性(模块)成本函数。此增强功能为强大的模型打开了大门,但同时也带来了非常困难的优化问题。在具有组合约束的子模最小化和在线子模最小化的最新进展的基础上,本演讲概述了具有组合约束的近似在线子模最小化的路线。我们以“最小合作切割”为例解决这种情况,其模块化模拟“最小切割”在许多应用中都很重要。

课程简介: Recently, there has been rising interest in replacing the linear (modular) cost function in combinatorial problems by a submodular cost function. This enhancement opens the door for powerful models, but at the same time leads to very hard optimization problems. Building on recent progress in submodular minimization with combinatorial constraints, and in online submodular minimization, this talk outlines routes for approximate online submodular minimization with combinatorial constraints. We address this scenario using the example of Minimum Cooperative Cut”, whose modular analogue, Minimum Cut, has been important in many applications.
关 键 词: 模块化模拟; 成本函数
课程来源: 视频讲座网
数据采集: 2020-12-29:zyk
最后编审: 2021-01-08:yumf
阅读次数: 46