贝叶斯网模型的学习、推理和应用

近年来在人工智能领域,不确定性问题一直成为人们关注和研究的焦点。贝叶斯网是用来表示不确定变量集合联合概率分布的图形模式,它反映了变量间潜在的依赖关系。使用贝叶斯网建模已成为解决许多不确定性问题的强有力工具。基于国内外最新的研究成果对贝叶斯网模型的学习、推理和应用情况进行了综述,并对未来的发展方向进行了展望。

一种大规模IP网络多链路拥塞推理算法

基于最小集覆盖理论的拥塞链路推理算法,仅对共享瓶颈链路进行推理,当拥塞路径存在多条链路拥塞时,算法的推理性能急剧下降.针对该问题,提出一种基于贝叶斯最大后验(Bayesian maximum a-posterior,简称BMAP)改进的拉格朗日松弛次梯度推理算法(Lagrange relaxation sub-gradient algorithm based on BMAP,简称LRSBMAP).针对推理算法中链路覆盖范围对算法推理性能的影响,以及探针部署及额外E2E路径探测发包的开销问题,提出设置度阈值(degree threshold value,简称DTV)参数预选待测IP网络收发包路由器节点,通过引入优选系数?,在保证链路覆盖范围的基础上,兼顾开销问题,确保算法的推理性能.针对大规模IP网络多链路拥塞场景下,链路先验概率求解方程组系数矩阵的稀疏性,提出一种对称逐次超松弛(symmetry successive over-relaxation,简称SSOR)分裂预处理共轭梯度法(preconditioned conjugate gradient method based on SSOR,简称PCG_SSOR)求解链路先验概率近似唯一解的方法,防止算法求解失败.实验验证了所提算法的准确性及鲁棒性.

BNs-OLS-SARIMA对城市短时交通流的预测

通过在目标路口构建贝叶斯交通网(BNs),并对与此交通网相关的交通流建立非平稳季节(SARIMA)模型,采用最小二乘法(OLS)取得相应模型的最佳权重组合,对缺失数据下的城市道路短时交通流进行预测。使用重庆市某路口的交通流数据对模型进行检测,通过多种预测指标对结果进行对比分析,结果表明BNs-OLS-SARIMA把交通流的网络结构与其周期性结合在一起,对短时交通流有良好的预测效果。

Development of Bayesian Network Models for Risk-Based Ship Design

In the past fifteen years, the attention of ship safety treatment as an objective rather than a constraint has started to sweep through the whole maritime industry. The risk-based ship design （RBD） methodology, advocating systematic integration of risk assessment within the conventional design process has started to takeoff. Despite this wide recognition and increasing popularity, important factors that could potentially undermine the quality of the results come from both quantitative and qualitative aspects during the risk assessment process. This paper details a promising solution by developing a formalized methodology for risk assessment through effective storing and processing of historical data combined with data generated through first-principle approaches. This method should help to generate appropriate risk models in the selected platform （Bayesian networks） which can be employed for decision making at design stare.

Predicting Complex Word Emotions and Topics through a Hierarchical Bayesian Network

In this paper, we provide a Word Emotion Topic （WET） model to predict the complex word e- motion information from text, and discover the dis- trbution of emotions among different topics. A complex emotion is defined as the combination of one or more singular emotions from following 8 basic emotion categories： joy, love, expectation, sur- prise, anxiety, sorrow, anger and hate. We use a hi- erarchical Bayesian network to model the emotions and topics in the text. Both the complex emotions and topics are drawn from raw texts, without con- sidering any complicated language features. Our ex- periment shows promising results of word emotion prediction, which outperforms the traditional parsing methods such as the Hidden Markov Model and the Conditional Random Fields（CRFs） on raw text. We also explore the topic distribution by examining the emotion topic variation in an emotion topic diagram.

A data-driven early micro-leakage detection and localization approach of hydraulic systems

Leakage is one of the most important reasons for failure of hydraulic systems.The accurate positioning of leakage is of great significance to ensure the safe and reliable operation of hydraulic systems.For early stage of leakage,the pressure of the hydraulic circuit does not change obviously and therefore cannot be monitored by pressure sensors.Meanwhile,the pressure of the hydraulic circuit changes frequently due to the influence of load and state of the switch,which further reduces the accuracy of leakage localization.In the work,a novel Bayesian networks(BNs)-based data-driven early leakage localization approach for multi-valve systems is proposed.Wavelet transform is used for signal noise reduction and BNs-based leak localization model is used to identify the location of leakage.A normalization model is developed to improve the robustness of the leakage localization model.A hydraulic system with eight valves is used to demonstrate the application of the proposed early micro-leakage detection and localization approach.

Using the contact network model and Metropolis-Hastings sampling to reconstruct the COVID-19 spread on the “Diamond Princess”

Traditional compartmental models such as SIR(susceptible,infected,recovered)assume that the epidemic transmits in a homogeneous population,but the real contact patterns in epidemics are heterogeneous.Employing a more realistic model that considers heterogeneous contact is consequently necessary.Here,we use a contact network to reconstruct unprotected,protected contact,and airborne spread to simulate the two-stages outbreak of COVID-19(coronavirus disease 2019)on the‘‘Diamond Princess"cruise ship.We employ Bayesian inference and Metropolis-Hastings sampling to estimate the model parameters and quantify the uncertainties by the ensemble simulation technique.During the early epidemic with intensive social contacts,the results reveal that the average transmissibility t was 0.026 and the basic reproductive number R0 was 6.94,triple that in the WHO report,indicating that all people would be infected in one month.The t and R0 decreased to 0.0007 and 0.2 when quarantine was implemented.The reconstruction suggests that diluting the airborne virus concentration in closed settings is useful in addition to isolation,and high-risk susceptible should follow rigorous prevention measures in case exposed.This study can provide useful implications for control and prevention measures for the other cruise ships and closed settings.