Inference Schemes for M Best Solutions for Soft CSPs
Inference Schemes for M Best Solutions for Soft CSPs
复制标题
软 CSP 的 M 最佳解决方案的推理方案
DOI：
--
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
R. Dechter
中科院分区：
文献类型：
--
作者：
E. Rollon;N. Flerova;R. Dechter
本文在半环的统一框架下给出了m-best任务的形式化。因此，已知的推理算法的定义和它们的正确性和完整性的m-最好的任务是立即暗示。我们还描述和分析了一个桶消除算法，用于解决m-最好的任务，elim-m-opt，在早期的研讨会上提出，并介绍了一个扩展的迷你桶框架，产生一个集合的边界为每个m-最好的解决方案。一些经验证明的算法和它们的潜力的近似。
The paper present a formalization of the m-best task within the unifying framework of semirings. As a consequence, known inference algorithms are defined and their correctness and completeness for the m-best task are immediately implied. We also describe and analyze a Bucket Elimination algorithm for solving the m-best task, elim-m-opt, presented in an earlier workshop and introduce an extension to the mini-bucket framework, yielding a collection of bounds for each of the m-best solutions. Some empirical demonstration of the algorithms and their potential for approximations are provided.