摘要
在专家系统中,利用条件概率进行推理存在着个关键问题:①逻辑与概率在表达推理规则的不一致;②难以进行高阶逻辑的推理.结合条件事件代数和Markov Chain Monte Carlo算法则提出了基于贝叶斯网的概率逻辑的推理方法.利用条件事件代数和条件事件,通过扩展普通的可测空间,首先使概率与逻辑在表达规则时相一致,然后通过条件事件代数把高阶条件事件转换成为普通的事件和逻辑事件的组合.通过使用Gibbs抽样算法对这些普通事件进行抽样达到稳态,最终估算出高阶条件事件的值完成高阶概率逻辑推理.通过一个实例说明了条件事件概率信息进行推理的方法.
For processing reasoning and resolving the discrepancy between logic and probability in making inferences from conditional probability information,we propose and implement a probabilistic logic reasoning approach on Bayesian networks,which combings Conditional Event Algebra and Markov Monte Carlo simulating algorithm.By extending normal measurable space with conditional event,we first bring logic consistent with probability in denoting conditional probability information,and then we transform a higher-order conditional event to normal events and correspond logical combination events via Conditional Event Algebra.We use Gibbs simulation to sample the normal events to be a stationary state.By computing the quantitative values of the events,we can evaluate the quantitative value of higher-order conditional event at last.An example of application of our method shows how we make inferences from conditional probability information.
出处
《云南大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第S2期308-312,共5页
Journal of Yunnan University(Natural Sciences Edition)
关键词
概率逻辑推理
条件事件代数
GIBBS抽样
probabilistic logic reasoning
conditional event algebra
gibbs sampler
