一种集成式不确定推理方法研究  

Research on an Ensemble Method of Uncertainty Reasoning

作　　者：贺怀清[1] 李建伏[1] 

机构地区：[1]中国民航大学计算机科学与技术学院,天津300300

出　　处：《计算机科学》2011年第6期246-250,共5页Computer Science

基　　金：国家自然科学基金项目(60879003);中央高校科研业务费(ZXH2009C001);中国民航大学引进人才科研启动费(09qd04S);天津市应用基础及前沿技术研究计划项目(10JCYBJC00900)资助

摘　　要：集成学习是采用某种规则把一系列学习器的结果进行整合以获得比单个学习器更好的学习效果的一种机器学习方法。研究表明集成学习是可行的,能取得比传统学习方法更好的性能。不确定推理是人工智能的重要研究方向之一,目前已经开发出了多种不确定推理方法,这些方法在实际应用中各有优缺点。借鉴集成学习,提出一种集成式不确定推理方法,其基本思想是按照一定的策略集成多种不确定推理方法,以提高推理的准确性。理论分析和实验结果验证了方法的合理性和可行性。Ensemble learning is a machine learning paradigm where multiple models are strategically generated and combined to obtain better predictive performance than a single learning method.It was proven that ensemble learning is feasible and tends to yield better results.Uncertainty reasoning is one of the important directions in artificial intelligence.Various uncertainty reasoning methods have been developed and all have their own advantages and disadvantages.Motivated by ensemble learning,an ensemble method of uncertainty reasoning was proposed.The main idea of the new methodis in accordance with the basic framework of ensemble learning,where multiple uncertainty reasoning methods is used in time and the result of various reasoning methods is integrated by some rules as the final result.Finally,theoretical analysis and experimental tests show that the ensemble uncertainty reasoning method is effective and feasible.

关 键 词：不确定推理 混合推理 集成学习 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]