马尔可夫逻辑网络（Markov logic network (MLN))是由一阶逻辑公式及其对应的权值组成的二元组集合。马尔可夫逻辑网络的基本思想是将一阶逻辑的限制放松，即一个事件违反公式越多，其发生概率越小，但未必为0。
马尔可夫逻辑网络（MLNs）提供了一种紧凑的语言来描述超大规模的马尔可夫网络，并能灵活地将各类领域知识模块化地整合其中。从一阶逻辑的角度看，MLNs具备可靠处理不确定性、容忍不完善与矛盾知识的能力，从而降低了系统脆弱性。统计关系学习中的诸多重要任务——如集体分类、链接预测、基于链接的聚类、社交网络建模及对象识别——均可自然转化为MLN学习与推理的具体实例。
參考文獻
*
外部連結
- [https://web.archive.org/web/20101011211905/http://ai.cs.washington.edu/projects/statistical-relational-learning University of Washington Statistical Relational Learning group]
- [http://alchemy.cs.washington.edu/ Alchemy 2.0: Markov logic networks in C++]
- [http://www.pracmln.org pracmln: Markov logic networks in Python]
- [http://ias.cs.tum.edu/probcog-wiki ProbCog: Markov logic networks in Python and Java that can use its own inference engine or Alchemy's]
- [https://web.archive.org/web/20080820073634/http://thebeast.googlecode.com/ markov thebeast: Markov logic networks in Java]
- [https://code.google.com/p/rockit/ RockIt: Markov logic networks in Java (with web interface/REST API)]
- [http://research.cs.wisc.edu/hazy/tuffy Tuffy: A Learning and Inference Engine with strong RDBMs-based optimization for scalability]
- [http://i.stanford.edu/hazy/felix/doc/ Felix: A successor to Tuffy, with prebuilt submodules to speed up common subtasks]
- [http://factorie.cs.umass.edu/usersguide/UsersGuide030Overview.html Factorie: Scala based probabilistic inference language, with prebuilt submodules for natural language processing etc]
- [https://www.cra.com/commercial-solutions/probabilistic-modeling-services.asp Figaro: Scala based MLN language]
- [https://github.com/anskarl/LoMRF LoMRF: Logical Markov Random Fields, an open-source implementation of Markov Logic Networks in Scala]
评论 (0)