Key points are not available for this paper at this time.
基于决策模型集成的贝叶斯平均(BA)允许评估决策的不确定性,这对于医疗诊断等安全关键应用至关重要。集成模型的可解释性还可以为负责做出可靠决策的专家提供有用信息。因此,决策树(DT)作为一种决策模型对专家具有吸引力。然而,对此类模型进行BA会使决策树集成模型变得不可解释。本文提出了一种贝叶斯决策树集成的概率解释新方法。该方法基于对决策树不确定性的量化评估,允许专家找到一个在预测准确性和结果可信度方面表现优异的决策树。为了使实验中对决策树的BA可行,我们采用了一种具有可逆跳跃扩展的马尔科夫链蒙特卡洛技术。临床数据的结果显示,就预测准确性而言,所提方法优于用于解释决策树集成的最大后验(MAP)方法。
Building similarity graph...
Analyzing shared references across papers
Loading...
Vitaly Schetinin
Jonathan E. Fieldsend
Derek Partridge
IEEE Transactions on Information Technology in Biomedicine
University of Exeter
Leicester Royal Infirmary
University of Bedfordshire
Building similarity graph...
Analyzing shared references across papers
Loading...
Schetinin 等人(星期二,)研究了该问题。
www.synapsesocial.com/papers/69e06826778f938530c11e72 — DOI: https://doi.org/10.1109/titb.2006.880553
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: