摘要
贝叶斯方法是概率统计学中一种很重要的方法。分类知识发现是数据挖掘的一项重要内容,研究各种高性能、高速度的分类算法是数据挖掘面临的主要问题之一。本文介绍了贝叶斯信念网络,并针对传统算法在时海量数据进行分类时速度较慢的缺点,提出了压缩候选的贝叶斯信念网络构造算法。它在不影响原有算法的可靠性的前提下,大大提高了学习速度,并通过在实际工作的执行情况来证明该算法的有效性。
Bayesian approach is an important method in statistics. Data classification is an important task of data miming. To discover a high-performance, high-speed classification is one of key problems for data mining. In this paper, we introduce the Bayesian belief nerwork. Because these algorithms are very slow,we introduce a method based on compressive candidates, which greatly speed up the study process. At last we prove that this method is reasonable {or its application on live data.
出处
《计算机科学》
CSCD
北大核心
2006年第9期157-158,共2页
Computer Science
基金
全国教育科学十五规划重点课题(No:AYA010034)基金资助。
关键词
贝叶斯网络
分类
数据挖掘
Bayesian belief network, Classification, Data mining
