Assessment of Catastrophic Risk Using Bayesian Network Constructed from Domain Knowledge and Spatial Data

Prediction of natural disasters and their consequences is difficult due to the uncertainties and complexity of multiple related factors. This article explores the use of domain knowledge and spatial data to construct a Bayesian network (BN) that facilitates the integration of multiple factors and quantification of uncertainties within a consistent system for assessment of catastrophic risk. A BN is chosen due to its advantages such as merging multiple source data and domain knowledge in a consistent system, learning from the data set, inference with missing data, and support of decision making. A key advantage of our methodology is the combination of domain knowledge and learning from the data to construct a robust network. To improve the assessment, we employ spatial data analysis and data mining to extend the training data set, select risk factors, and fine‐tune the network. Another major advantage of our methodology is the integration of an optimal discretizer, informative feature selector, learners, search strategies for local topologies, and Bayesian model averaging. These techniques all contribute to a robust prediction of risk probability of natural disasters. In the flood disaster's study, our methodology achieved a better probability of detection of high risk, a better precision, and a better ROC area compared with other methods, using both cross‐validation and prediction of catastrophic risk based on historic data. Our results suggest that BN is a good alternative for risk assessment and as a decision tool in the management of catastrophic risk.     

A Neural Network-Based Method for Risk Factor Analysis of West Nile Virus

There is a lack of knowledge about which risk factors are more important in West Nile virus (WNV) transmission and risk magnitude. A better understanding of the risk factors is of great help in developing effective new technologies and appropriate prevention strategies for WNV infection. A contribution analysis of all risk factors in WNV infection would identify those major risk factors. Based on the identified major risk factors, measures to control WNV proliferation could be directed toward those significant risk factors, thus improving the effectiveness and efficiency in developing WNV control and prevention strategies. Neural networks have many generally accepted advantages over conventional analytical techniques, for instance, ability to automatically learn the relationship between the inputs and outputs from training data, powerful generalization ability, and capability of handling nonlinear interactions. In this article, a neural network model was developed for analysis of risk factors in WNV infection. To reveal the relative contribution of the input variables, the neural network was trained using an algorithm called structural learning with forgetting. During the learning, weak neural connections are forced to fade away while a skeletal network with strong connections emerges. The significant risk factors can be identified by analyzing this skeletal network. The proposed approach is tested with the dead bird surveillance data in Ontario, Canada. The results demonstrate the effectiveness of the proposed approach.     

基于GMM-RBF神经网络的前列腺癌诊断方法

 前列腺癌是近年来发病率上升速度最快的男性癌症，严重威胁着患者的身体健康，准确地判断癌症患者的患病情况对于节约医疗资源、提高患者满意度起着至关重要的作用。近年来，基于数据挖掘的癌症诊断方法逐渐成为疾病诊断领域的研究热点，在提高诊断准确性上显示出极大优势。        针对现有前列腺癌早期诊断方法准确性不高的问题，提出一种基于高斯混合模型改进径向基函数神经网络的前列腺癌诊断方法--GMM-RBF神经网络方法。该方法通过使用高斯混合模型对径向基函数神经网络中径向基函数的参数进行预训练，使模型避免陷入局部最优，之后采用改进的粒子群优化算法对神经网络进行训练。采用国家临床医学科学数据中心提供的数据进行前列腺癌诊断实验，将所提出的方法与径向基神经网络、分类回归树、支持向量机和逻辑回归等主流的机器学习算法进行对比，并使用准确性、特异性、敏感性和AUC值对模型的性能进行评价。        研究结果表明，与改进前的神经网络模型相比，GMM-RBF神经网络模型收敛速度更快、初始准确度更高；与其它机器学习算法相比，GMM-RBF神经网络模型在10折交叉验证中取得了较高的准确性、敏感性、特异性和AUC值。        GMM-RBF神经网络方法在模型预测精度上比传统的径向基函数神经网络模型有很大提升，能够得到更为可靠的前列腺癌诊断结果，为医疗工作者初步诊断前列腺癌和穿刺活检操作提供有效的辅助决策支持，该方法的提出对于减少患者痛苦、提高患者满意度和节约医疗资源具有实际意义。     

基于云模型关联规则的企业转型战略风险预警

针对评价指标数据的特点,构造了一种基于云模型的数值型关联规则挖掘算法,并将其运用于企业转型战略风险预警。首先运用云模型约简评价指标;然后,采用属性空间软划分方法对定量型属性的定义域进行划分,使定量型关联规则挖掘转换为定性关联规则挖掘,此基础上提取规则模版;最后采用有规则约束的Apriori算法挖掘云关联规则,并对检验样本风险等级进行判别。实证分析结果表明,与标准BP神经网络模型相比,该模型是一种更为有效和实用的战略风险预警工具。     

A national risk analysis model (NRAM) for the assessment of COVID-19 epidemic

COVID-19 has caused a critical health concern and severe economic crisis worldwide. With multiple variants, the epidemic has triggered waves of mass transmission for nearly 3 years. In order to coordinate epidemic control and economic development, it is important to support decision-making on precautions or prevention measures based on the risk analysis for different countries. This study proposes a national risk analysis model (NRAM) combining Bayesian network (BN) with other methods. The model is built and applied through three steps. (1) The key factors affecting the epidemic spreading are identified to form the nodes of BN. Then, each node can be assigned state values after data collection and analysis. (2) The model (NRAM) will be built through the determination of the structure and parameters of the network based on some integrated methods. (3) The model will be applied to scenario deduction and sensitivity analysis to support decision-making in the context of COVID-19. Through the comparison with other models, NRAM shows better performance in the assessment of spreading risk at different countries. Moreover, the model reveals that the higher education level and stricter government measures can achieve better epidemic prevention and control effects. This study provides a new insight into the prevention and control of COVID-19 at the national level.     

A Bayesian Method to Mine Spatial Data Sets to Evaluate the Vulnerability of Human Beings to Catastrophic Risk

Vulnerability of human beings exposed to a catastrophic disaster is affected by multiple factors that include hazard intensity, environment, and individual characteristics. The traditional approach to vulnerability assessment, based on the aggregate‐area method and unsupervised learning, cannot incorporate spatial information; thus, vulnerability can be only roughly assessed. In this article, we propose Bayesian network (BN) and spatial analysis techniques to mine spatial data sets to evaluate the vulnerability of human beings. In our approach, spatial analysis is leveraged to preprocess the data; for example, kernel density analysis (KDA) and accumulative road cost surface modeling (ARCSM) are employed to quantify the influence of geofeatures on vulnerability and relate such influence to spatial distance. The knowledge‐ and data‐based BN provides a consistent platform to integrate a variety of factors, including those extracted by KDA and ARCSM to model vulnerability uncertainty. We also consider the model's uncertainty and use the Bayesian model average and Occam's Window to average the multiple models obtained by our approach to robust prediction of the risk and vulnerability. We compare our approach with other probabilistic models in the case study of seismic risk and conclude that our approach is a good means to mining spatial data sets for evaluating vulnerability.     

基于顾客需求偏好的面向无人零售终端的供应链资源整合优化研究

面向新零售的无人零售终端所关注的焦点是如何为顾客提供满意的购物体验。目前,无人零售终端在提供服务时较多关注不同场景下顾客对商品类型的需求,而往往忽略价格、品质和终端操作便捷性等其他差异化的需求。而无人零售的无人值守、需要灵活补货和配货等特点,决定了其在提供差异化服务时,离不开供应链网络的有效支撑,这就需要对供应链网络中的各类资源进行合理整合。本文在分析不同场景下顾客需求偏好的基础上,重点探讨了面向无人零售终端资源整合中的供应链成员筛选机理,有针对性地提出了关键因素的挖掘和分类方法。在此基础上,构建了对应的整合优化模型和求解算法,并通过算例验证了方法和算法的可行性、有效性与可操作性。     

Estimation of distribution algorithms using Gaussian Bayesian networks to solve industrial optimization problems constrained by environment variables

Many real-world optimization problems involve two different subsets of variables: decision variables, and those variables which are not present in the cost function but constrain the solutions, and thus, must be considered during optimization. Thus, dependencies between and within both subsets of variables must be considered. In this paper, an estimation of distribution algorithm (EDA) is implemented to solve this type of complex optimization problems. A Gaussian Bayesian network is used to build an abstraction model of the search space in each iteration to identify patterns among the variables. As the algorithm is initialized from data, we introduce a new hyper-parameter to control the influence of the initial data in the decisions made during the EDA execution. The results show that our algorithm improves the cost function more than the expert knowledge does.

     

The min-p robust optimization approach for facility location problem under uncertainty

Improper value of the parameter p in robust constraints will result in no feasible solutions while applying stochastic p-robustness optimization approach (p-SRO) to solving facility location problems under uncertainty. Aiming at finding the lowest critical p-value of parameter p and corresponding robust optimal solution, we developed a novel robust optimization approach named as min-p robust optimization approach (min-pRO) for P-median problem (PMP) and fixed cost P-median problem (FPMP). Combined with the nearest allocation strategy, the vertex substitution heuristic algorithm is improved and the influencing factors of the lowest critical p-value are analyzed. The effectiveness and performance of the proposed approach are verified by numerical examples. The results show that the fluctuation range of data is positively correlated with the lowest critical p-value with given number of new facilities. However, the number of new facilities has a different impact on lowest critical p-value with the given fluctuation range of data. As the number of new facilities increases, the lowest critical p-value for PMP and FPMP increases and decreases, respectively.

     

具有最优学习率的RBF神经网络及其应用

传统固定学习率的RBF神经网络在金融时间序列预测方面已经有比较成功的应用,但网络学习率的选择问题却给传统RBF神经网络的使用带来了不便.利用梯度下降法及优化方法推导出了RBF神经网络的动态最优学习率并将其应用于网络学习算法,具有最优学习率的RBF神经网络能够在保证网络稳定学习的同时兼顾网络的收敛速度.为了检验具有动态最优学习率的RBF神经网络的预测效果,对沪深300指数波动率进行了预测实验.实验结果表明,具有动态最优学习率的RBF神经网络比传统的固定学习率的RBF神经网络有着更快的收敛速度,同时也避免了人为选定学习率的不便.     

一种改进CLOSET算法

CLOSET算法是一种挖掘频繁闭项集的有效算法。本文分析了CLOSET算法在闭合性检查以及剪枝效率方面的一些不足,并针对这些问题,引入共生项集的概念,利用共生项集的一些性质,提出了一种基于共生项集的改进算法。     

基于领域知识和聚类的关联规则深层知识发现研究

本文针对传统关联规则挖掘算法产生大量冗余规则,提出了对关联规则结果进行二次挖掘,并设计了算法对挖掘出的关联规则进行聚类,然后基于已有领域知识对聚类后的关联规则进行新颖度评价,对于新颖度较高价值较大的关联规则可以存储于领域知识库用于决策使用或再次挖掘过程。该算法有效的减少的规则的数量,提高了规则的新颖性和精确度,对商业应用具有很高的价值。文章最后使用UCI开源数据进行了实验分析,并验证了该算法的有效性。     

基于相似日聚类和贝叶斯神经网络的光伏发电功率预测研究

光伏发电功率的预测是光伏发电规划和运行的基础,因而受到越来越多的重视。文中提出了FCM相似日聚类与智能算法相结合的光伏阵列功率短期预测模型。该方法的思路是首先通过分析影响光伏阵列输出功率的主要因素,对历史数据与预测日气象环境进行模糊分类,并筛选出相似度高的子集作为样本,以提高预测样本的质量;然后通过神经网络映射出特征空间与光伏功率之间的复杂关系,并用贝叶斯理论对神经网络参数进行优化,提高网络的泛化能力。为检验该方法的有效性和精确性,将所提出方法与常用BP神经网络模型对同一仿真算例进行预测,预测结果表明本文提出的预测模型效果更佳。     

重要数据跨境流动背景下风险路径的识别与分级

重要数据的跨境流动引发了数据安全、国家安全等风险挑战。风险路径的识别和分级是对重要数据跨境流动进行预警管理的重要内容。本文基于复杂网络中的二分网络模型,对重要数据的跨境流动进行研究。首先,通过重要数据跨境流动的二分网络和关联网络识别风险路径;其次,构建基于网络结构和接收节点属性的目标风险路径方法以计算其风险值;最后,对我国某重要行业跨境流动的数据开展实证分析,验证算法的有效性和精准度。本文旨在为重要数据跨境流动的预警管理提供量化方法,有效预防重要数据跨境流动带来的风险,提升我国数据治理能力。     

An efficient spread-based evolutionary algorithm for solving dynamic multi-objective optimization problems

Dynamic multi-objective optimization algorithms are used as powerful methods for solving many problems worldwide. Diversity, convergence, and adaptation to environment changes are three of the most important factors that dynamic multi-objective optimization algorithms try to improve. These factors are functions of exploration, exploitation, selection and adaptation operators. Thus, effective operators should be employed to achieve a robust dynamic optimization algorithm. The algorithm presented in this study is known as spread-based dynamic multi-objective algorithm (SBDMOA) that uses bi-directional mutation and convex crossover operators to exploit and explore the search space. The selection operator of the proposed algorithm is inspired by the spread metric to maximize diversity. When the environment changed, the proposed algorithm removes the dominated solutions and mutated all the non-dominated solutions for adaptation to the new environment. Then the selection operator is used to select desirable solutions from the population of non-dominated and mutated solutions. Generational distance, spread, and hypervolume metrics are employed to evaluate the convergence and diversity of solutions. The overall performance of the proposed algorithm is evaluated and investigated on FDA, DMOP, JY, and the heating optimization problem, by comparing it with the DNSGAII, MOEA/D-SV, DBOEA, KPEA, D-MOPSO, KT-DMOEA, Tr-DMOEA and PBDMO algorithms. Empirical results demonstrate the superiority of the proposed algorithm in comparison to other state-of-the-art algorithms.

     

An efficient scheduling algorithm for the hot coil making in the steel mini-mill

This paper investigates and suggests an efficient solution to the problem of scheduling the steel making line in the Mini Steel Mill, which consists of three major processes: molten steel making, continuous slab casting, and hot charged rolling. Careful synchronization of these processes is a key productivity factor, since a very limited amount of work-in-process inventory is allowed. Since each process must run in batch, the schedule for the Mini-Mill consists of grouping and sequencing of lots for each process. However, each process has its own criteria for judging the quality of its lot grouping, which often conflicts with other processes. An efficient scheduling algorithm for the Mini-Mill is proposed. Numerical experiments with real world data suggest that the proposed algorithm yield satisfactory schedules very efficiently. The algorithm is currently used for the actual scheduling of a Mini-Mill in Korea.     

A sustainable supply chain network considering lot sizing with quantity discounts under disruption risks: centralized and decentralized models

This study proposes a framework for the main parties of a sustainable supply chain network considering lot-sizing impact with quantity discounts under disruption risk among the first studies. The proposed problem differs from most studies considering supplier selection and order allocation in this area. First, regarding the concept of the triple bottom line, total cost, environmental emissions, and job opportunities are considered to cover the criteria of sustainability. Second, the application of this supply chain network is transformer production. Third, applying an economic order quantity model lets our model have a smart inventory plan to control the uncertainties. Most significantly, we present both centralized and decentralized optimization models to cope with the considered problem. The proposed centralized model focuses on pricing and inventory decisions of a supply chain network with a focus on supplier selection and order allocation parts. This model is formulated by a scenario-based stochastic mixed-integer non-linear programming approach. Our second model focuses on the competition of suppliers based on the price of products with regard to sustainability. In this regard, a Stackelberg game model is developed. Based on this comparison, we can see that the sum of the costs for both levels is lower than the cost without the bi-level approach. However, the computational time for the bi-level approach is more than for the centralized model. This means that the proposed optimization model can better solve our problem to achieve a better solution than the centralized optimization model. However, obtaining this better answer also requires more processing time. To address both optimization models, a hybrid bio-inspired metaheuristic as the hybrid of imperialist competitive algorithm (ICA) and particle swarm optimization (PSO) is utilized. The proposed algorithm is compared with its individuals. All employed optimizers have been tuned by the Taguchi method and validated by an exact solver in small sizes. Numerical results show that striking similarities are observed between the results of the algorithms, but the standard deviations of PSO and ICA–PSO show better behavior. Furthermore, while PSO consumes less time among the metaheuristics, the proposed hybrid metaheuristic named ICA–PSO shows more time computations in all small instances. Finally, the provided results confirm the efficiency and the performance of the proposed framework and the proposed hybrid metaheuristic algorithm.

     

一个新的银行信用风险评估方法

本文以银行信用风险管理为例,将粗糙集和决策树两种具有互补优势的数据挖掘方法相结合,对客户信用做出归类分析判断,最后利用决策树生成决策规则.实践证明,这种方法忠于原始数据,提高了分类准确度,减小了决策树规模,具有良好的性能.     

基于Dijkstra-PD-ACO算法的大城市公交线路优化与评价方法研究

城市公交线路的布设受到诸多影响因素的作用,单纯的追求线路单一指标的"最优"在实际公交线路优化时往往难以取得满意的效果。首先针对传统公交线路优化算法在求解线路优化问题中存在的不足,基于蚁群优化算法的寻优特性,结合Dijkstra算法在局部路径寻优中的优点,提出了Dijkstra蚁群混合优化算法。其次对于线路优化所得到的可行备选方案集,基于分层聚类主成分分析评价法进行优化效能评价。最后以合肥市一环内公交线路优化为实例进行验证。结果表明, 本文提出的方法在兼顾客流密度最大、出行路径最短的同时给出了有效公交线路备选方案,优化结果也符合合肥市的实际情况。同时,该方法对我国大中城市公交线网优化问题的研究具有一定的借鉴意义。