Bayesian网的独立性推广模型

The Extension of the Bayesian Network Based on Independence

本文提出了Bayesian网的独立性推广模型。Bayesian网能够表示变量之间概率形响关系与条件独立性,但不能表示因果独立性。虽然Noisy OR模型能够较好地表示变量之间的因果独立性,但该模型又因只能表示因果独立性而具有很大的局限性。本文提出的独立性推广模型解决了Bayesian网因果独立性表示能力不足的问题,扩展了Bayesian网与Noisy OR模型的表示范围,同时简化了Bayesian网的条件概率表,并且新模型更能够反映变量之间的概率影响关系。实验结果表明了该模型的实用性。

The Bayesian can express the conditionally independence conveniently,but in appliactions it can't handle causally independence easily. The Noisy OR model can express causally independence well,but the limitation of the Noisy OR model blocks the widely use of itself. In this paper,we present the extension of the Bayesian Network based on independence. Our new model generalizes the Bayesian Networks to handle causally independence properly,and simplifies the conditionally probability table. Experimental results show the availability of our new model.