自动化学习出贝叶斯网络的图形结构是机器学习领域的一个挑战。其基本思想可以追溯到由 Rebane 和 Pearl 提出的恢复算法。<ref>{{cite book | vauthors = Rebane G, Pearl J | chapter = The Recovery of Causal Poly-trees from Statistical Data| title = Proceedings, 3rd Workshop on Uncertainty in AI | location = Seattle, WA | pages = 222–228 | year = 1987 | arxiv = 1304.2736}}</ref>该算法的基础是三节点有向无环图中的三种可能模式: | 自动化学习出贝叶斯网络的图形结构是机器学习领域的一个挑战。其基本思想可以追溯到由 Rebane 和 Pearl 提出的恢复算法。<ref>{{cite book | vauthors = Rebane G, Pearl J | chapter = The Recovery of Causal Poly-trees from Statistical Data| title = Proceedings, 3rd Workshop on Uncertainty in AI | location = Seattle, WA | pages = 222–228 | year = 1987 | arxiv = 1304.2736}}</ref>该算法的基础是三节点有向无环图中的三种可能模式: |