Sensitivity analysis in discrete Bayesian networks

This paper presents an efficient computational method for performing sensitivity analysis in discrete Bayesian networks. The method exploits the structure of conditional probabilities of a target node given the evidence. First, the set of parameters which is relevant to the calculation of the conditional probabilities of the target node is identified. Next, this set is reduced by removing those combinations of the parameters which either contradict the available evidence or are incompatible. Finally, using the canonical components associated with the resulting subset of parameters, the desired conditional probabilities are obtained. In this way, an important saving in the calculations is achieved. The proposed method can also be used to compute exact upper and lower bounds for the conditional probabilities, hence a sensitivity analysis can be easily performed. Examples are used to illustrate the proposed methodology     

Linking Bayesian networks and PLS path modeling for causal analysis

Causal knowledge based on causal analysis can advance the quality of decision-making and thereby facilitate a process of transforming strategic objectives into effective actions. Several creditable studies have emphasized the usefulness of causal analysis techniques. Partial least squares (PLS) path modeling is one of several popular causal analysis techniques. However, one difficulty often faced when we commence research is that the causal direction is unknown due to the lack of background knowledge. To solve this difficulty, this paper proposes a method that links the Bayesian network and PLS path modeling for causal analysis. An empirical study is presented to illustrate the application of the proposed method. Based on the findings of this study, conclusions and implications for management are discussed.     

Exploiting causal independence in large Bayesian networks

The assessment of a probability distribution associated with a Bayesian network is a challenging task, even if its topology is sparse. Special probability distributions based on the notion of causal independence have therefore been proposed, as these allow defining a probability distribution in terms of Boolean combinations of local distributions. However, for very large networks even this approach becomes infeasible: in Bayesian networks which need to model a large number of interactions among causal mechanisms, such as in fields like genetics or immunology, it is necessary to further reduce the number of parameters that need to be assessed. In this paper, we propose using equivalence classes of binomial distributions as a means to define very large Bayesian networks. We analyse the behaviours obtained by using different symmetric Boolean functions with these probability distributions as a means to model joint interactions. Some surprisingly complicated behaviours are obtained in this fashion, and their intuitive basis is examined.     

基于因果语义定向的贝叶斯网络结构学习

基于变量之间基本依赖关系、基本结构、d-separation标准、依赖分析思想和混合定向策略,给出了一种有效实用的贝叶斯网络结构学习方法,不需要结点有序,并能避免打分-搜索方法存在的指数复杂性,以及现有依赖分析方法的大量高维条件概率计算等问题。     

基于因果影响独立模型的贝叶斯网络参数学习

基于因果影响独立模型及其中形成的特定上下文独立关系,提出一种适于样本学习的贝叶斯网络参数学习算法。该算法在对局部概率模型降维分解的基础上,通过单父节点条件下的子节点概率分布来合成局部结构的条件概率分布,参数定义复杂度较低且能较好地处理稀疏结构样本集。实验结果表明,该算法与标准最大似然估计算法相比,能充分利用样本信息,具有较好的学习精度。     

基于贝叶斯网络的系统可靠性分析平台

贝叶斯网络是一种进行不确定性推理和分析的有效工具,针对系统可靠性分析问题,建立了一种基于贝叶斯网络的系统可靠性分析平台。所建立的分析平台将贝叶斯网络应用于系统可靠性分析中,把系统各组部件抽象成节点,从而构成贝叶斯网络模型;通过推理和分析算法找到影响系统可靠性的薄弱环节;通过计算某个部件发生变化时对整个系统的影响,得出各个节点的重要度,给出合理的优化方案,提升系统可靠性。通过对平视显示器实例进行分析,计算了每个组部件对整个系统的影响程度,证明该分析平台在利用贝叶斯网络对系统可靠性分析上的可行性。     

基于模糊贝叶斯网的危害性分析方法

针对传统的故障模式、影响与危害性分析(FMECA)方法不足的问题,提出了一个基于模糊贝叶斯网的危害性分析方法。该方法将模糊理论与贝叶斯网推理技术结合起来,用三角模糊数来描述专家的模糊评分值;通过模糊集合映射,将其转化为评级的模糊子集;以置信结构的模糊规则,表示故障模式的属性与危害度之间的关系;利用贝叶斯网络推理算法综合置信结构的模糊规则,通过贝叶斯网推理得到模糊子集形式的危害度,再经过去模糊计算,得到故障危害等级的清晰值,从而确定故障模式的危害程度。实验结果表明,所提方法能够提高传统分析方法的准确性和应用范围。     

用于因果分析的混合贝叶斯网络结构学习

目前主要结合扩展的熵离散化方法和打分一搜索方法进行混合贝叶斯网络结构学习，算法效率和可靠性低，而且易于陷入局部最优结构。针对问题建立了一种新的混合贝叶斯网络结构迭代学习方法．在迭代中，基于父结点结构和Gibbs sampling进行混合数据聚类，实现对连续变量的离散化，再结合贝叶斯网络结构优化调整，使贝叶斯网络结构序列逐渐趋于稳定，可避免使用扩展的熵离散化和打分——搜索所带来的主要问题．     

Bayesian MLP neural networks for image analysis

We demonstrate the advantages of using Bayesian multi-layer perceptron (MLP) neural networks for image analysis. The Bayesian approach provides consistent way to do inference by combining the evidence from the data to prior knowledge from the problem. A practical problem with MLPs is to select the correct complexity for the model, i.e., the right number of hidden units or correct regularization parameters. The Bayesian approach offers efficient tools for avoiding overfitting even with very complex models, and facilitates estimation of the confidence intervals of the results. In this contribution we review the Bayesian methods for MLPs and present comparison results from two case studies. In the first case, MLPs were used to solve the inverse problem in electrical impedance tomography. The Bayesian MLP provided consistently better results than other methods. In the second case, the goal was to locate trunks of trees in forest scenes. With Bayesian MLP it was possible to use large number of potentially useful features and prior for determining the relevance of the features automatically.     

Comparative analysis of discretization methods in Bayesian networks

A key step in implementing Bayesian networks (BNs) is the discretization of continuous variables. There are several mathematical methods for constructing discrete distributions, the implications of which on the resulting model has not been discussed in literature. Discretization invariably results in loss of information, and both the discretization method and the number of intervals determines the level of such loss. We designed an experiment to evaluate the impact of commonly used discretization methods and number of intervals on the developed BNs. The conditional probability tables, model predictions, and management recommendations were compared and shown to be different among models. However, none of the models did uniformly well in all comparison criteria. As we cannot justify using one discretization method against others, we recommend caution when discretization is used, and a verification process that includes evaluating alternative methods to ensure that the conclusions are not an artifact of the discretization approach.     

An acceptability estimation and analysis methodology based on Bayesian networks

As companies are forced to conceive innovative products to stay competitive, designers face the challenge of developing products more suited to users' needs and perceptions in order to be accepted, thus reducing project risk failure. Evaluating users' acceptability has become an important research problem. Current approaches leave the acceptance evaluation question to be answered in the last stages of product development process (NPD), when an almost finished prototype is available and when there is no time left for important modifications. Acceptability evaluation methods suitable for use from the early stages of the NPD process are thus needed. This paper proposes a method for acceptability evaluation and analysis that can be used in the early stages of the development cycle. It is based on the evaluation of the solution concept by the users. The relationships among the factors (or criteria) are made explicit, thus helping designers to identify the key factors for acceptance. As the users' tests and the maturity of the concept prototype are limited in this stage, the proposed method exploits the inference properties of Bayesian networks making it possible to make useful estimations and allowing the exploration of actions that could improve the product acceptability level. Two case studies are presented in order to illustrate the method, the first related to a technological product design for a home-health care service provider and the second to a work-related musculoskeletal disorder prevention software design.Relevance to industryThe article describes an acceptability assessment and an analysis approach to be used by industrial engineers, designers and ergonomists in the early phases of design projects. The method can help the design team to identify the levers (key factors) for enhancing product acceptance and to identify different actions (e.g. product modification, deployment strategy, and training).     

基于故障贝叶斯网的冷链物流系统可靠性分析

借鉴可靠性工程理论中系统可靠性的分析方法,将冷链物流系统运行故障这一抽象问题具体化处理,根据系统中各功能环节的运行特点及事件之间的因果关系,构建冷链物流系统的系统失效故障树;在此基础上生成贝叶斯网络,以综合评估冷链物流系统的运行可靠性,揭示系统故障产生的主要原因,为改进冷链物流系统的运行可靠性提供定量依据.将所提出方法用于某第三方冷链物流企业的运作管理中,取得了预期效果.     

贝叶斯网络在学生模型建模中的应用分析

将课程教学资源融合到学生模型构建中,描述了包括领域知识拓扑结构的建立、条件概率表学习算法的推理的详细过程,最终得到了学生模型中关于章节知识项的贝叶斯网络结构图,并通过一个实验系统对个性化教学系统中学生模型建构的整个框架的可行性进行了验证.     

基于因果效应的贝叶斯网络结构学习方法

从数据中学习贝叶斯网络结构是一个NP-hard问题,提高网络结构学习算法精度是研究的重难点。基于Judea Pearl因果理论,提出了一种贝叶斯网络结构学习方法,提升了现有算法的准确率。利用改进的Pearl因果效应和BDe评分,学习网络节点优先次序,利用K2算法学习初始网络,并通过BDe评分反向调节、互信息和BDe评分删除边以修正学习结果。实验在贝叶斯网络标准数据集ASIA、ALARM上进行,在样本量为2000~20000的20组实验中,学习准确率较MMHC算法平均提升16%,准确率标准差较MMHC算法平均缩小17%。实验表明,基于因果效应的方法较MMHC算法有更好地性能。     

基于函数逼近的多层前馈神经网络灵敏度分析

神经网络灵敏度分析对网络结构设计、硬件实现等具有重要的指导意义,已有的灵敏度计算公式对权值和输入扰动有一定限制或者计算误差较大。基于Piché的随机模型,通过使用两个逼近函数对神经网络一类Sigmoid激活函数进行高精度逼近,获得了新的神经网络灵敏度计算公式,公式取消了对权值扰动和输入扰动的限制,与其他方法相比提高了计算精度,实验证明了公式的正确性和精确性。     

Sensitivity analysis of single hidden-layer neural networks withthreshold functions

An important consideration when applying neural networks to pattern recognition is the sensitivity to weight perturbation or to input errors. In this paper, we analyze the sensitivity of single hidden-layer networks with threshold functions. In a case of weight perturbation or input errors, the probability of inversion error for an output neuron is derived as a function of the trained weights, the input pattern, and the variance of weight perturbation or the bit error probability of the input pattern. The derived results are verified with a simulation of the Madaline recognizing handwritten digits. The result shows that the sensitivity of trained networks is far different from that of networks with random weights.     

连续航班延误与波及的贝叶斯网络分析

针对空运系统日益严重的航班延误,尝试将贝叶斯方法应用于航班数据分析,重点考虑同一飞机飞行连续航班的情况。借助Netica软件包,建立贝叶斯网络模型。通过贝叶斯网络推理进行连续航班延误波及分析,并用实际航班数据进行测试。结果表明,概率统计意义下,模型能够清晰反映连续航班延误原因分布、过站时间差分布和按时间段的延误波及情况。     

Competitiveness analysis of automotive industry in Turkey using Bayesian networks

The purpose of this study is to analyze the relations between the factors that enable national competitive advantage and the establishment of competitive superiority in automotive industry through a comprehensive analytical model. Bayesian networks (BN) are used to investigate the associations of different factors in the automotive industry which lead to competitive advantage. The results of the study focus on building a road map for the automotive sector policy makers in their way to improve the competitiveness through scenario analysis. Using the probabilistic dependency structure of the Bayesian network all of the variables in the model can be estimated. Thus, with the proposed model the automotive industry can be analyzed as a whole system and not only in terms of single variables. Findings of the model indicate that technological developments in automotive industry can alter the nature of competition in this industry.     

Bayesian networks precipitation model based on hidden Markov analysis and its application

Surface precipitation estimation is very important in hydrologic forecast. To account for the influence of the neighbors on the precipitation of an arbitrary grid in the network, Bayesian networks and Markov random field were adopted to estimate surface precipitation. Spherical coordinates and the expectation-maximization (EM) algorithm were used for region interpolation, and for estimation of the precipitation of arbitrary point in the region. Surface precipitation estimation of seven precipitation station...     

Bayesian learning of Bayesian networks with informative priors

This paper presents and evaluates an approach to Bayesian model averaging where the models are Bayesian nets (BNs). A comprehensive study of the literature on structural priors for BNs is conducted. A number of prior distributions are defined using stochastic logic programs and the MCMC Metropolis-Hastings algorithm is used to (approximately) sample from the posterior. We use proposals which are tightly coupled to the priors which give rise to cheaply computable acceptance probabilities. Experiments using data generated from known BNs have been conducted to evaluate the method. The experiments used 6 different BNs and varied: the structural prior, the parameter prior, the Metropolis-Hasting proposal and the data size. Each experiment was repeated three times with different random seeds to test the robustness of the MCMC-produced results. Our results show that with effective priors (i) robust results are produced and (ii) informative priors improve results significantly.