基本信息
浏览量:46
职业迁徙
个人简介
Graphical models have become an indispensible tool in machine learning and applied statistics for representing networks of variables and probability distributions describing their interactions. Recovering the maximum a posteriori (MAP) configuration of random variables in a graphical model is an important problem with applications ranging from protein folding to image processing. The task of finding the optimal MAP configuration has been shown to be NP-hard in general (Shimony, 1994), and has been well studied from several different perspectives, including message-passing algorithms (Pearl, 1988), linear programming relaxations (Taskar et al., 2004), and max-flow/min-cut frameworks (Greig et al., 1989). This work uses recent advances in the theory of perfect graphs to outline situations where provably-optimal solutions are possible to find with efficient algorithms in P, and provides several useful applications of this result.
Consider an undirected graph, G=(V,E) with n vertices V={v1,...,vn} and edges E: V×V$ →{0,1}. Such a graph is perfect if every induced subgraph has chromatic number equal to its clique number. Perfect graphs originated with work by Claude Berge (1963); a graph is Berge if it and its complement contains no odd holes (i.e., no chordless cycles of length 5, 7, 9,..., etc.). It is fairly straightforward to show that all perfect graphs are Berge, but the converse is not obvious; work in (Chudnovsky et al., 2006) enumerates the entire family of Berge graphs and shows that all family members are perfect.
Consider an undirected graph, G=(V,E) with n vertices V={v1,...,vn} and edges E: V×V$ →{0,1}. Such a graph is perfect if every induced subgraph has chromatic number equal to its clique number. Perfect graphs originated with work by Claude Berge (1963); a graph is Berge if it and its complement contains no odd holes (i.e., no chordless cycles of length 5, 7, 9,..., etc.). It is fairly straightforward to show that all perfect graphs are Berge, but the converse is not obvious; work in (Chudnovsky et al., 2006) enumerates the entire family of Berge graphs and shows that all family members are perfect.
研究兴趣
论文共 12 篇作者统计合作学者相似作者
按年份排序按引用量排序主题筛选期刊级别筛选合作者筛选合作机构筛选
时间
引用量
主题
期刊级别
合作者
合作机构
msra(2009)
引用25浏览0引用
25
0
BIOINFORMATICSno. 1 (2005): i144-51
加载更多
作者统计
合作学者
合作机构
D-Core
- 合作者
- 学生
- 导师
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn