Nodes In Bayesian Network Research Articles

BAYESIAN NETWORK FOR RISK ASSESSMENT OF ABRASION PROCESSES IN THE COASTAL AREA OF THE SEA OF AZOV

The coast of the Sea of Azov is characterized by a high population density and economic development. At the same time, a significant part of the sea coast is subject to abrasion. High coastal cliff retreat rates determine the relevance of assessing possible losses in this region. The aim of the work is to develop a model for assessing the risk of abrasion processes in the coastal zone of the Sea of Azov. Bayesian networks, which are often used to study coastal processes, were selected as a model toolkit. To assess the abrasion hazard in the coastal zone of the Sea of Azov, a network, consisting of two subnets, is proposed. The “Hazard Assessment” subnet describes the effect of exogenous factors on the abrasion rate. The “Risk Assessment” subnet is designed to determine the consequences (the magnitude of expected losses) of the abrasion process. The main attention is paid to the characteristics of the Bayesian network nodes. In the model, the risk from hazardous coastal processes is expressed in natural terms: the lost land area and the number of damaged facilities located on the coast. The Bayesian model is coupled with a geographic information system on base of the geospatial representation of the study region. An example of assessment the lost land area for a part of the Taganrog Bay coast is considered. Comparison of the losses estimates based on Bayesian network and average abrasion rate is given. Proposed probabilistic method provide additional information, enriching the decision-making process.

Dynamic Knowledge Inference Based on Bayesian Network Learning

On the basis of studying datasets of students' course scores, we constructed a Bayesian network and undertook probabilistic inference analysis. We selected six requisite courses in computer science as Bayesian network nodes. We determined the order of the nodes based on expert knowledge. Using 356 datasets, the K2 algorithm learned the Bayesian network structure. Then, we used maximum a posteriori probability estimation to learn the parameters. After constructing the Bayesian network, we used the message-passing algorithm to predict and infer the results. Finally, the results of dynamic knowledge inference were presented through a detailed inference process. In the absence of any evidence node information, the probability of passing other courses was calculated. A mathematics course (a basic professional course) was chosen as the evidence node to dynamically infer the probability of passing other courses. Over time, the probability of passing other courses greatly improved, and the inference results were consistent with the actual values and can thus be visualized and applied to an actual school management system.

A Novel BN Learning Algorithm Based on Block Learning Strategy.

Learning accurate Bayesian Network (BN) structures of high-dimensional and sparse data is difficult because of high computation complexity. To learn the accurate structure for high-dimensional and sparse data faster, this paper adopts a divide and conquer strategy and proposes a block learning algorithm with a mutual information based K-means algorithm (BLMKM algorithm). This method utilizes an improved K-means algorithm to block the nodes in BN and a maximum minimum parents and children (MMPC) algorithm to obtain the whole skeleton of BN and find possible graph structures based on separated blocks. Then, a pruned dynamic programming algorithm is performed sequentially for all possible graph structures to get possible BNs and find the best BN by scoring function. Experiments show that for high-dimensional and sparse data, the BLMKM algorithm can achieve the same accuracy in a reasonable time compared with non-blocking classical learning algorithms. Compared to the existing block learning algorithms, the BLMKM algorithm has a time advantage on the basis of ensuring accuracy. The analysis of the real radar effect mechanism dataset proves that BLMKM algorithm can quickly establish a global and accurate causality model to find the cause of interference, predict the detecting result, and guide the parameters optimization. BLMKM algorithm is efficient for BN learning and has practical application value.

Mapping Flood-Based Farming Systems with Bayesian Networks

Many actors in agricultural research, development, and policy arenas require accurate information on the spatial extents of cropping and farming practices. While remote sensing provides ways for obtaining such information, it is often difficult to distinguish between different types of agricultural practices or identify particular farming systems. Stochastic system behavior or similarity in the spectral signatures of different system components can lead to misclassification. We addressed this challenge by using a probabilistic reasoning engine informed by expert knowledge and remote sensing data to map flood-based farming systems (FBFS) across Kisumu County in Kenya and the Tigray region in Ethiopia. Flood-based farming is an important form of agricultural production employed in regions with seasonal water surplus, which can be harvested and used to irrigate crops. Geographic settings for FBFS vary widely in terms of hydrology, vegetation, and local practices of agronomic flooding. Agronomic success is often difficult to anticipate, because the timing and amount of flooding usually cannot be precisely predicted. We generated a Bayesian network model to describe the FBFS settings of the study regions. We acquired three years (2014–2016) of Moderate Resolution Imaging Spectroradiometer (MODIS) Terra spectral data as eight-day composite time series and elevation data from the Shuttle Radar Topography Mission (SRTM) to compute 10 spatial data metrics corresponding to 10 of the 17 Bayesian network nodes. We used the spatial data metrics in a fully probabilistic framework to generate the 10 spatial data nodes. We then used these as inputs for the probabilistic model to generate prior and posterior spatial estimates for specific metrics along with their spatially explicit uncertainties. We show how such an approach can be used to predict plausible areas for FBFS based on several scenarios. We demonstrate how spatially explicit information can be derived from remote sensing data as fuzzy quantifiers for incorporating uncertainties when mapping complex systems. The approach achieved a remarkably accurate result in both study areas, where 84–90% of various FBFS fields sampled were correctly mapped as having a high chance of being suitable for the practice.

Railway dangerous goods transportation system risk analysis: An Interpretive Structural Modeling and Bayesian Network combining approach

In this paper, an Interpretive Structural Modeling (ISM) and Bayesian Network (BN) combining approach is applied to analyze the relationships and interaction strengths among the risk factors or accident causes of railway dangerous goods transportation system (RDGTS), quantitatively. According to the statistical data, 17 sub-indicators are concluded and applied as the nodes in BN. Based on the 17 sub-indicators and with the help of experts, the combined ISM and BN approach is used to analyze the system risk with seven steps: establish the relation matrix, calculate the reachability matrix, divide the reachability matrix into different levels, form the directed graph, form the Bayesian Network, establish the prior marginal and conditional probabilities of the sub-indicators in the BN, causal reasoning for the BN and obtain the final probabilities of occurrence for all sub-indicators. The conditional probability tables (CPTs) are used to express the relationships and interaction strengths among the nodes and the parent node variables. The final analysis results show that the sub-indicator, staffs are lack of technical and knowledge during transportation processes, has the highest impact to the RDGTS with probability 0.74, and the sudden natural disaster has the weakest impact to the RDGTS with probability 0.05.

Selecting data adaptive learner from multiple deep learners using Bayesian networks

A method to predict time-series using multiple deep learners and a Bayesian network is proposed. In this study, the input explanatory variables are Bayesian network nodes that are associated with learners. Training data are divided using K-means clustering, and multiple deep learners are trained depending on the cluster. A Bayesian network is used to determine which deep learner is in charge of predicting a time-series. We determine a threshold value and select learners with a posterior probability equal to or greater than the threshold value, which could facilitate more robust prediction. The proposed method is applied to financial time-series data, and the predicted results for the Nikkei 225 index are demonstrated.

Risk analysis for reservoir flood control operation considering two-dimensional uncertainties based on Bayesian network

The uncertainty of initial water level being above the designed flood limited water level and the uncertainty of inflow caused by flood forecast error can cause flood risk in real-time reservoir flood operation. A risk analysis model for reservoir flood regulation under two-dimensional uncertainties based on Bayesian network is proposed. The nodes of Bayesian network are determined and the network structure is established with expert knowledge; the parameter learning is conducted with the training samples obtained from Monte Carlo simulation. Thereafter, through the prior probability inference without posterior information and the posterior probability inference with given posterior information, the variation of flood risk is analyzed under singular-factor uncertainty and two-factors uncertainties. The case study of Xianghongdian Reservoir using a flood of 100 years return period indicates: the risk resulted from inflow uncertainty is greater than that of the uncertainty of initial water level; there is a certain complementarity between the uncertainties of inflow and initial water level, and the combination risk is between the results of two single-factor risk levels. Moreover, Bayesian Network is able to conduct bi-directional inferences and infer the probability distribution of any other node, which has practical value for risk assessment and control of reservoir flood control operation.

Fire risk assessment in cotton storage based on fuzzy comprehensive evaluation and Bayesian network

SummaryCotton storage fire develops rapidly, and it has high probability of being out of control and causing huge economic loss. In order to carry out a quantitative risk assessment of cotton storage fire accidents, a dynamic risk assessment model of cotton storage fire is established by means of data analysis, fuzzy comprehensive evaluation (FCE), cases study and expert opinions based on Bayesian network (BN). The model includes three sections (fire causes section, fire discoverer section, fire control section) and one mainline, with 22 nodes, and gives accident risk in the form of economic loss. Risk can be renewed quickly by giving new evidence to model nodes of BN. The application of FCE provides a feasible solution to solve the data deficiency problem in databases, experiment results and experience in model construction. Sensitivity analysis result shows firefighters participation, timely control of fire, firefighting systems and workers are important to put out the fire. A case study shows that the loss of a fire accident happened in an open storage yard is more serious than the loss of a similar fire accident happened in warehouse. This model provides support for cotton storage fire risk management and decision‐making in an emergency fire accident.

Cause Analysis of Unsafe Behaviors in Hazardous Chemical Accidents: Combined with HFACs and Bayesian Network.

Hazardous chemical accidents (HCAs) seriously endanger public life, property, and health. Human and organizational factors are important causes of many kinds of accidents. In order to systematically explore the influencing factors of unsafe behaviors in HCAs in China, the method of human factors analysis and classification system based on the Bayesian network (BN-HFACs) was introduced. According to the 39 investigation reports of HCAs in China, the origin Bayesian network (BN) was obtained and the failure sensitivity of every node in BN was calculated. The results have shown that hazardous material environment (1.63) and mechanical equipment (0.49) in the level of preconditions of unsafe behavior have the same direction failure effect with operation error, while there is no factor has the same direction failure effect with operation violate. Some factors in organization influence and unsafe supervision, such as organization climate (0.34), operation guidance (0.37), planned operation (0.22), and legal supervision (0.19), are also important reasons for operational errors, while resource management (0.12), hidden investigation (0.18) and legal supervision (0.13) have an impact on operation violates. Moreover, there are still close relationships between other hierarchical elements, such as the operation guidance effect on the hazardous material environment (6.60), and the organizational climate has the most obvious impact on other factors at the level of organizational factors. Based on the above research conclusions, suggestions for individual, enterprise, and government were put forward, respectively, and the limitations of this study were also clarified.

EBNO: Evolution of cost‐sensitive Bayesian networks

AbstractThe last decade has seen an increase in the attention paid to the development of cost‐sensitive learning algorithms that aim to minimize misclassification costs while still maintaining accuracy. Most of this attention has been on cost‐sensitive decision tree learning, whereas relatively little attention has been paid to assess if it is possible to develop better cost‐sensitive classifiers based on Bayesian networks. Hence, this paper presents EBNO, an algorithm that utilizes Genetic algorithms to learn cost‐sensitive Bayesian networks, where genes are utilized to represent the links between the nodes in Bayesian networks and the expected cost is used as a fitness function. An empirical comparison of the new algorithm has been carried out with respect to (a) an algorithm that induces cost‐insensitive Bayesian networks to provide a base line, (b) ICET, a well‐known algorithm that uses Genetic algorithms to induce cost‐sensitive decision trees, (c) use of MetaCost to induce cost‐sensitive Bayesian networks via bagging (d) use of AdaBoost to induce cost‐sensitive Bayesian networks, and (e) use of XGBoost, a gradient boosting algorithm, to induce cost‐sensitive decision trees. An empirical evaluation on 28 data sets reveals that EBNO performs well in comparison with the algorithms that produce single interpretable models and performs just as well as algorithms that use bagging and boosting methods.

Real-time risk analysis of road tanker containing flammable liquid based on fuzzy Bayesian network

The risk of road transportation of flammable liquid is of great uncertainty due to the time-varying conditions of the passing locations and the environment, which leads to challenging risk analysis. In this work, a real-time risk analysis method for road tanker transportation based on the fuzzy Bayesian network (FBN) is proposed. The bow-tie model is first employed to identify hazards in flammable liquid road transportation systems and shows risk evolution. The framework of the Bayesian network (BN) is then determined accordingly. In the case that the historical statistics of accidents are limited, a probabilistic estimation model that combines expert judgment and fuzzy set theory is established to determine the prior probabilities and the conditional probabilities of the BN nodes. Case studies of typical road tanker transportation accidents were carried out to show the risk level variation with both the internal and external conditions at different moments. Sensitivities of the parent nodes were analyzed, and the critical factors leading to accidents were identified. Studies show that this method can dynamically characterize the changes in both the probabilities and the consequence levels of road tanker transport accidents. Based on the vehicle’s GPS data and the local environment, the proposed method can provide an estimation of the real-time risk for road tankers.

Logic‐based probabilistic network model to detect and track faults in a process system

AbstractProcess systems are becoming complex due to a higher dependency among operational variables and complex control loops. Principal component analysis (PCA) is widely used to reduce the dimensionality of the complex process systems, while Bayesian networks (BNs) are increasingly employed to model relationships among the operational variables. This article integrates these two methods (BN and PCA) through a logic‐based approach to study the fault conditions of a process system. A distillation pilot plant is used to test the integrated approach. The process monitoring data are analyzed using the PCA to identify the abnormality variables while the BN is developed using data‐driven learning. The variable dependency in the BN nodes is learned through maximum likelihood estimation. The results of the proposed approach are compared against the logic‐based full BN model. The study observes that the logic‐based PCA‐BN approach proposed improves the reliability of fault detection. While the logic‐based full BN provides a better understanding of fault propagation path through the unit, which helps track and troubleshoot the detected faults.

The relationship determination between the Bayesian networks nodes based on the intuitionistic fuzzy set

The relationship determination between the Bayesian networks nodes based on the intuitionistic fuzzy set

Modeling Incomplete Knowledge of Semantic Web Using Bayesian Networks

ABSTRACTInteroperable ontologies already exist in the biomedical field, enabling scientists to communicate with minimum ambiguity. Unfortunately, ontology languages, in the semantic web, such as OWL and RDF(S), are based on crisp logic and thus they cannot handle uncertain knowledge about an application field, which is unsuitable for the medical domain. In this paper, we focus on modeling incomplete knowledge in the classical OWL ontologies, using Bayesian networks, all keeping the semantic of the first ontology, and applying algorithms dedicated to learn parameters of Bayesian networks in order to generate the Bayesian networks. We use EM algorithm for learning conditional probability tables of different nodes of Bayesian network automatically, contrary to different tools of Bayesian networks where probabilities are inserted manually. To validate our work, we have applied our model on the diagnosis of liver cancer using classical ontology containing incomplete instances, in order to handle medical uncertain knowledge, for predicting a liver cancer.

A Bayesian network model of lymphatic tumor progression for personalized elective CTV definition in head and neck cancers

Many tumors including head and neck squamous cell carcinoma (HNSCC) spread along the lymphatic network. Current imaging modalities can only detect sufficiently large metastases. Therefore, adjacent lymph node levels (LNL) are irradiated electively since they may harbor microscopic tumors. We apply Bayesian Networks (BN) to model lymphatic tumor progression. The model can subsequently be used to personalize the risk estimation of microscopic lymph node metastases in newly diagnosed patients based on their distribution of macroscopic metastases. A BN is a graphical representation of a joint probability distribution. We represent LNLs by binary random variables corresponding to the BN nodes. Each LNL is associated with a hidden microscopic state and an observed macroscopic state (e.g. 18F-FDG-PET/CT imaging). The primary tumor is represented by network input nodes. We demonstrate the concept for early T-stage oropharyngeal carcinomas and their spread to ipsilateral lymph node levels (LNL) Ib to IV. We show that the BN parameters can be efficiently learnt by merging pathology findings on microscopic tumor progression (which is limited to a few published studies) and imaging data on macroscopic tumor progression such as CT and 18F-FDG-PET (which are widely available in clinical practice). The trained network can be used to quantify how the distribution of macroscopic metastases impacts the probability of microscopic involvement of the remaining LNLs. The analysis suggests that the risk of microscopic involvement of level IV exceeds 5% only if level III harbors metastases. Excluding level IV from the elective CTV for other patients would reduce the integral dose delivered to the patient and potentially reduce acute and late side effects.

Inversion of rice canopy chlorophyll content and leaf area index based on coupling of radiative transfer and Bayesian network models

The radiative transfer model (RTM) simulates forward spectral reflectance of vegetation and is used to estimate physical parameters using backwards inversion. However, differentiation of spectral reflectances may be hampered due to model parameter combinations, and the cost function within RTM that calculates statistical distance may lead to inconsistent inversions. Bayesian network (BN) is a probabilistic model that is used to solve problems of model ambiguity and incompleteness. Here, we constructed a model to estimate rice growth parameters using data collected by an unmanned aerial vehicle (UAV). We collected rice canopy spectral information using a MiniMCA-6 multispectral camera fitted to an UAV that was used to determine BN structure using parameters derived from the PROSAIL model. We calculated conditional probability distributions of different observed combinations of rice canopy chlorophyll content (CCC) and leaf area index (LAI) and a look up table of maximum conditional probabilities of rice growth parameters based on BN was developed. Results indicated that most accurate inversions of LAI and CCC as BN nodes were achieved at reflectances of 720 nm, 800 nm under the red normalized difference vegetation index and at reflectances of 550, 720, 800 nm under the modified simple ratio index, respectively. Compared with the cost function inversion method, the BN method mitigated the ill-posed problem of inversion and obtained higher inversion accuracy with model test R2, RRMSE, and RE values of 0.81, 0.31, and 0.38, and 0.83, 0.36, and 0.43 for LAI and CCC, respectively. We conclude that application of the BN method to the inversion process of crop RTM could improve inversion accuracy of estimation of crop parameters.

Matrix-based Bayesian Network for efficient memory storage and flexible inference

For real-world civil infrastructure systems that consist of a large number of functionally and statistically dependent components, such as transportation systems or water distribution networks, the Bayesian Network (BN) can be a powerful tool for probabilistic inference. In a BN, the statistical relationship between multiple random variables (r.v.’s) is modeled through a directed acyclic graph. The complexity of inference in the BN depends not only on the number of r.v.’s, but also the graphical structure. As a consequence, the application of standard BN techniques may become infeasible even with a moderate number of r.v.’s as the size of an event set exponentially increases with the number of r.v.’s. Moreover, when the exhaustive set that is required for full quantification of a discrete BN node becomes intractably large, only approximate inference algorithms are feasible, which do not require the full (explicit) description of all BN nodes. We address both issues in discrete BNs by proposing a matrix-based Bayesian Network (MBN) that facilitates efficient modeling of joint probability mass functions and flexible inference. The MBN is developed for exact as well as approximate BN inference. The efficiency and applicability of the MBN are demonstrated by numerical examples. The supporting source code and data are available for download at https://github.com/jieunbyun/GitHub-MBN-code.

Exploiting Bayesian networks for fault isolation: A diagnostic case study of diesel fuel injection system

Fault isolation is known to be a challenging problem in machinery troubleshooting. It is not only because the isolation of multiple faults contains considerable number of uncertainties due to the strong correlation and coupling between different faults, but often massive prior knowledge is needed as well. This paper presents a Bayesian network-based approach for fault isolation in the presence of the uncertainties. Various faults and symptoms are parameterized using state variables, or the so-called nodes in Bayesian networks (BNs). Probabilistically causality between a fault and a symptom and its quantization are described respectively by a directed edge and conditional probability. To reduce the qualitative and quantitative knowledge needed, particular considerations are given to the simplification of Bayesian networks structures and conditional probability expressions using rough sets and noisy-OR/MAX model, respectively. By adopting the simplified approach, symptoms under multiple-fault are decoupled into the ones under every single fault, while the quantity of the conditional probabilities is simplified into the linear form of the faults quantity. Prior knowledge needed in Bayesian network-based diagnostic model is reduced significantly, which decreases the complexity in establishing and applying this diagnosis model. The computational efficiency is improved accordingly in the simplified BN model, after eliminating the redundant symptoms. The fault isolation methodology is illustrated through an example of diesel engine fuel injection system to verify the developed model.

Risk Assessment of Underground Subway Stations to Fire Disasters Using Bayesian Network

Subway station fires often have serious consequences because of the high density of people and limited number of exits in a relatively enclosed space. In this study, a comprehensive model based on Bayesian network (BN) and the Delphi method is established for the rapid and dynamic assessment of the fire evolution process, and consequences, in underground subway stations. Based on the case studies of typical subway station fire accidents, 28 BN nodes are proposed to represent the evolution process of subway station fires, from causes to consequences. Based on expert knowledge and consistency processing by the Delphi method, the conditional probabilities of child BN nodes are determined. The BN model can quantitatively evaluate the factors influencing fire causes, fire proof/intervention measures, and fire consequences. The results show that the framework, combined with Bayesian network and the Delphi method, is a reliable tool for dynamic assessment of subway station fires. This study could offer insights to a more realistic analysis for emergency decision-making on fire disaster reduction, since the proposed approach could take into account the conditional dependency in the fire propagation process and incorporate fire proof/intervention measures, which is helpful for resilience and sustainability promotion of underground facilities.

Bayesian analysis of school bus accidents: a case study of China

Prevention and control of school bus accidents have been a hot spot topic around the world. The catastrophic accident can result in severe casualties associated with negative social impacts. In this study, a Bayesian network (BN) model is established for school bus accident assessments considering the influential factors including human error, vehicle failure, environmental impacts and management deficiency. The conditional probabilities of a few root nodes are statistically obtained based on school bus accidents that occurred in the past decade in China. The conditional probabilities of other BN nodes are determined by expert knowledge with treatment by the Dempster–Shafer evidence theory. The consequences of different scenarios of school accidents are estimated via changing the state values of some BN nodes. Furthermore, by conducting sensitivity analysis to the proposed BN, it is identified that “overload” is the most influential factor causing a school accident. The results of the proposed model indicate that the integration of Bayesian network and the Dempster–Shafer evidence theory is an effective framework for school bus accident assessment, which could provide more practical analysis for school bus accidents. This study could contribute to providing technical supports for school bus safety particularly in developing countries.