|
At present, the methods of learning the hidden variables of Bayesian network with known structure is mainly for Bayesian networks with discrete variables or Gaussian networks with continuous variables. In this paper, the method of learning the hidden variables of hybrid Bayesian network with discrete and continuous variables is presented. The discretization of continuous variables is not needed. The hidden variables are found by prior knowledge or the dimension of cliques... At present, the methods of learning the hidden variables of Bayesian network with known structure is mainly for Bayesian networks with discrete variables or Gaussian networks with continuous variables. In this paper, the method of learning the hidden variables of hybrid Bayesian network with discrete and continuous variables is presented. The discretization of continuous variables is not needed. The hidden variables are found by prior knowledge or the dimension of cliques in the moral graph of Bayesian network. The values of hidden variable are made based on dependency structure (star structure or prior structure) between variables and Gibbs sampling. The optimum dimension of hidden variable is made by combining extended MDL criterion with statistics method. Experimental results show that this method can effectively learn the hidden variables of hybrid Bayesian network with known structure. 目前,具有已知结构的隐藏变量学习主要针对具有离散变量的贝叶斯网和具有连续变量的高斯网.该文给出了具有连续和离散变量的混合贝叶斯网络隐藏变量学习方法.该方法不需要离散化连续变量,依据专业知识或贝叶斯网络道德图中Cliques的维数发现隐藏变量的位置,基于依赖结构(星形结构或先验结构)和Gibbs抽样确定隐藏变量的值,结合扩展的MDL标准和统计方法发现隐藏变量的最优维数.实验结果表明,这种方法能够有效地进行具有已知结构的混合贝叶斯网络隐藏变量学习. Bayesian network theory was used to model opponent's plan in a multi-robot confrontation system.A hybrid Bayesian network was constructed to identify the opponent's plan.Whole game field for soccer robots will be divided into some small areas and the Bayesian network analyzes and identifies to which area the opponent will kick the ball to implement final antagonism target of the soccer robot system.A strategy simulation system based on the opponent's plan modeling was constructed... Bayesian network theory was used to model opponent's plan in a multi-robot confrontation system.A hybrid Bayesian network was constructed to identify the opponent's plan.Whole game field for soccer robots will be divided into some small areas and the Bayesian network analyzes and identifies to which area the opponent will kick the ball to implement final antagonism target of the soccer robot system.A strategy simulation system based on the opponent's plan modeling was constructed and experiment results show it efficient. 用贝叶斯网络来解决多机器人对抗系统的对手建模问题,建立了用于一类多机器人对抗系统对手规划识别的混合贝叶斯网络。将足球机器人赛场进行分区,使用贝叶斯网络来分析和判断对手的意图为将球踢向哪个分区,实现足球机器人系统的对抗目标。建立了基于对手建模的策略仿真系统,实验结果表明了该策略仿真系统的有效性。 A modified fuzzy Bayesian network(FBN) is proposed in this study.It uses Gaussian mixture models(GMM) to make a fuzzy procedure for continuous image features.This particular procedure will transform continuous variables into discrete ones by soft quantizers,when dealing with continuous inputs with probabilistic and uncertain nature.It builds a hybrid Bayesian network(BN) construction modeling the causality of image features and diseases with expert knowledge,and trains the BN with data through... A modified fuzzy Bayesian network(FBN) is proposed in this study.It uses Gaussian mixture models(GMM) to make a fuzzy procedure for continuous image features.This particular procedure will transform continuous variables into discrete ones by soft quantizers,when dealing with continuous inputs with probabilistic and uncertain nature.It builds a hybrid Bayesian network(BN) construction modeling the causality of image features and diseases with expert knowledge,and trains the BN with data through machine learning,and estimates a probability of diseases by probability inference.This method is applied in prediction of the astrocytoma malignant degree and achieves an accuracy of 83.33%,which outperforms the BN using a crisp quantizer by a k-nearest neighbor classifier.This model provides more reasonable knowledge expression for domains with fuzzy and uncertain nature and a novel objective intelligent method to quantitatively assess the astrocytoma malignant level that can be used to assist doctors to diagnose the tumor. 针对医学影像特征具有模糊性和不确定性的特点,提出一种基于模糊贝叶斯网络的影像诊断预测模型。该模型使用高斯混合模型(GMM)对连续的视觉特征进行模糊量化处理,利用专家知识根据病症与影像特征之间的因果关系建立混合贝叶斯网络结构;由数据通过机器学习确定网络参数;采用概率推理定量估计病症的发生概率,从而建立一个可计算的预测模型。将该方法应用于星形细胞瘤分级预测,实验结果得出83.33%的正确识别率,远远超过使用最小近邻分类器(K-NN)实现连续变量硬(crisp)量化的贝叶斯网络模型,更合理地表达了具有模糊性、不确定性的专业领域的结构性知识,为星形细胞瘤恶性程度预测提供了新的辅助手段。
|