基础知识
条件概率(Conditional Probability)
![](https://img-blog.csdn.net/20141207195333153?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvYXNwaXJpbnZhZ3JhbnQ=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/SouthEast)
![](https://img-blog.csdn.net/20141207195353417?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvYXNwaXJpbnZhZ3JhbnQ=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/SouthEast)
相互独立时,p(A | B) = p(A)
贝叶斯规则
贝叶斯网络(Bayesian Network)定了一个独立的结构:一个节点的概率仅依赖于它的父节点。贝叶斯网络适用于稀疏模型,即大部分节点之间不存在任何直接的依赖关系。
![](https://img-blog.csdn.net/20141207195548436?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvYXNwaXJpbnZhZ3JhbnQ=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/SouthEast)
![](https://img-blog.csdn.net/20141207195603100?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvYXNwaXJpbnZhZ3JhbnQ=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/SouthEast)
联合概率(Joint Probability),表示所有节点共同发生的概率,将所有条件概率相乘:
![](https://img-blog.csdn.net/20141207195618311?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvYXNwaXJpbnZhZ3JhbnQ=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/SouthEast)
我们最终的目标是计算准确的边缘概率(Marginal Probability),比如计算Hangover的概率,边缘概率为各种状态下所有其他节点对本节点影响的概率的和。
边缘概率(Marginal Probability):即