| 增强型朴素贝叶斯产 |
| |
| 引用本文： | 王实,高文.增强型朴素贝叶斯产[J].计算机科学,2000,27(4):46-49. |
| |
| 作者姓名： | 王实 高文 |
| |
| 作者单位： | 中国科学院计算技术研究所,北京,100080 |
| |
| 摘 要： | 朴素贝叶斯是一种分类监督学习方法。在理论上,应用其前提为例子的属性值独立于例子的分类属性。这个前提在实际应用中过于严格,常常得不到满足,即使是这样,在违反该前提的情况下,朴素贝叶斯学习方法仍然取得了很大的成功。近来,一种改进的朴素贝叶斯方法,增强(Boost-ing),受到广泛的关注,AdaBoost方法是其主要方法。当AdaBoost方法被用于联合几个朴素贝叶斯分类器时,其在数学上等价于一个具有稀疏编码输入,单隐层节点,sigmoid激活函数的反馈型神经网络。
|
| 关 键 词： | 神经网络 朴素贝叶斯学习 学习算法 |
Boosting Naive Bayesian Learning |
| |
| Abstract: | In this paper we introduce the Boosting Naive Bayesian Learning.The Naive Bayesian classifier provides a simple approach to discriminate classification learning that has demonstrated competitive predictive accuracy on many learning tasks and it was broadly used in Data Mining field.Boosting combines the weak hypotheses such as Naive Bayesian by summing their probabilistic predictions.On the real world datasets,Boosting Naive Bayesian method has generalized perfect performance than the best published result using any other learning method. |
| |
| Keywords: | Data mining Naive Bayesian Boosting |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
点击此处可从《计算机科学》下载全文 |
|
