概率逻辑模型与学习研究进展

Advances in Probabilistic Logic Model and Learning

近年来,概率逻辑学习研究取得了很大进展,已经提出各种不同的形式化方法和学习方法,包括概率关系模型(PRMs)、贝叶斯逻辑程序(BLPs)、逻辑贝叶斯网络(LBNs)和随机逻辑程序(SLPs)等。文章重点介绍了贝叶斯网络与一阶逻辑的结合,并以PRMs、BLPs和LBNs为例,描述了基于贝叶斯网络的概率逻辑模型(PLMs)的知识表示方法,给出了此类PLMs一般使用的参数估计方法和结构学习方法,并给出了建议的研究方向。

Probabilistic logic learning （PLL）research has made significant progress over the last years. A rich variety of different formalisms and learning techniques have been developed, including probabilistic relational models, bayesian logic programs, and logic bayesian networks and stochastic logic programs etc. This paper, focusing on the combination of bayesian networks and first-order logic, provides an introductory survey on probabilistic logic models based on bayesian networks through the investigation of knowledge representation, parameter estimation and structure learning algorithms. Although the PLL community has successfully demonstrated the feasibility of a number of probabilistic models for relational data, there is much work on efficiency and scalability to be done in order to begin generalizing the range and applicability of the various models.