Dynamic Bayesian Network Framework Research Articles

A condition-informed dynamic Bayesian network framework to support severe accident management in nuclear power plants

In the nuclear industry, Severe Accident Management Guidelines (SAMGs) are developed with the objective of providing prescriptive actions for mitigating the consequences of Nuclear Power Plant (NPP) accidents. These guidelines are defined with respect to a set of expert-based prototypical severe accident scenarios and are symptom-based, i.e., they relate to the values of specific monitored plant parameters. Expert-based SAMGs might fail to deal with non-prototypical scenarios, i.e., scenarios not considered a priori by the experts. To address this issue, in this paper we propose a methodology based on Dynamic Bayesian Networks (DBNs) to support operators decision-making for avoiding the escalation of severe accidents to non-prototypical scenarios. The methodology is based on the integration of condition monitoring data into a DBN to allow it to i) infer the current Plant Operating Condition (POC), Initiating Event (IE), and NPP safety barriers Damage State (DS), ii) predict the grace time (GT) available to carry out mitigative actions, and iii) evaluate the effectiveness of the alternative candidate mitigative actions taking into account the safety barriers estimated DS and the predicted available GT. The proposed methodology is exemplified on a Loss of Coolant Accident (LOCA) in a WWER-1000.

Pyramid Dynamic Bayesian Networks for Key Performance Indicator Prediction in Long Time-Delay Industrial Processes

Building Dynamic Bayesian Networks (DBN) for time-delay industrial processes has always been tough, since the structure learning of DBN is a NP-hard problem. In this article, a pyramid DBN framework is proposed to speed up modeling and improve feature learning for industrial processes with large time delays. In the pyramid DBN framework, a sequence of small-sized DBNs are established, each of which is a progressively simpler representation of the previous layer, and the feature information learned by the DBN sequence corresponds to the layers in the pyramid. With information fusion of the feature pyramid and a further feature filtering with hill climbing based on Bayesian information criterion, we can restore the feature information of the time-delay industrial processes. Regression models are built based on the features learned by the pyramid DBN framework for further Key Performance Indicator (KPI) estimation. Advantages of the method have been effectively validated on two actual industrial cases. With multi-level feature learning and filtering, both modeling speed and prediction accuracy have been greatly improved. The proposed method outperforms other similar works on identifying important features in industrial processes and existing DBN-based soft sensors.

A Kalman Variational Autoencoder Model assisted by Odometric Clustering for Video Frame Prediction and Anomaly Detection.

The combination of different sensory information to predict upcoming situations is an innate capability of intelligent beings. Consequently, various studies in the Artificial Intelligence field are currently being conducted to transfer this ability to artificial systems. Autonomous vehicles can particularly benefit from the combination of multi-modal information from the different sensors of the agent. This paper proposes a method for video-frame prediction that leverages odometric data. It can then serve as a basis for anomaly detection. A Dynamic Bayesian Network framework is adopted, combined with the use of Deep Learning methods to learn an appropriate latent space. First, a Markov Jump Particle Filter is built over the odometric data. This odometry model comprises a set of clusters. As a second step, the video model is learned. It is composed of a Kalman Variational Autoencoder modified to leverage the odometry clusters for focusing its learning attention on features related to the dynamic tasks that the vehicle is performing. We call the obtained overall model Cluster-Guided Kalman Variational Autoencoder. Evaluation is conducted using data from a car moving in a closed environment [1] and leveraging a part of the University of Alcalá DriveSet dataset [2], where several drivers move in a normal and drowsy way along a secondary road.

Dynamic Bayesian network for durability of reinforced concrete structures in long-term environmental exposures

Reinforced concrete (RC) structures under the marine environment may be subjected to chloride-induced corrosion of reinforcement, which significantly impacts the structural serviceability and reliability and further affects the sustainability and development of society. However, most of the existing durability assessment methods for RC structures only address their static and deterministic durability prediction and assessment at the design stage given the constant environment, ignoring the influences of stochastic environmental effects, uncertainties in structural properties, and inspection results. To this end, this paper proposes a dynamic Bayesian network (DBN) based durability assessment framework combined with a deterioration model that considers random changes in environmental parameters, convective chloride ion transport, and corrosion-induced cracking of concrete. In this framework, two-dimensional chloride transport and its influences on the durability deterioration assessment are concerned and achieved using the finite difference method. Besides, to reduce the deviations in probabilistic evaluation, the good-lattice-point-set-partially stratified-sampling (GLP-PSS) method is employed to establish a DBN framework. The proposed DBN framework is used for sensitivity analysis through a real-world example to examine the effects of the environmental model, chloride transport mode, and inspection results of concrete crack on durability assessment.

Learning dynamic Bayesian networks from time-dependent and time-independent data: Unraveling disease progression in Amyotrophic Lateral Sclerosis

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease causing patients to quickly lose motor neurons. The disease is characterized by a fast functional impairment and ventilatory decline, leading most patients to die from respiratory failure. To estimate when patients should get ventilatory support, it is helpful to adequately profile the disease progression. For this purpose, we use dynamic Bayesian networks (DBNs), a machine learning model, that graphically represents the conditional dependencies among variables. However, the standard DBN framework only includes dynamic (time-dependent) variables, while most ALS datasets have dynamic and static (time-independent) observations. Therefore, we propose the sdtDBN framework, which learns optimal DBNs with static and dynamic variables. Besides learning DBNs from data, with polynomial-time complexity in the number of variables, the proposed framework enables the user to insert prior knowledge and to make inference in the learned DBNs. We use sdtDBNs to study the progression of 1214 patients from a Portuguese ALS dataset. First, we predict the values of every functional indicator in the patients’ consultations, achieving results competitive with state-of-the-art studies. Then, we determine the influence of each variable in patients’ decline before and after getting ventilatory support. This insightful information can lead clinicians to pay particular attention to specific variables when evaluating the patients, thus improving prognosis. The case study with ALS shows that sdtDBNs are a promising predictive and descriptive tool, which can also be applied to assess the progression of other diseases, given time-dependent and time-independent clinical observations.

Quantum Fluctuation Theorems beyond Two-Point Measurements.

We derive detailed and integral quantum fluctuation theorems for heat exchange in a quantum correlated bipartite thermal system using the framework of dynamic Bayesian networks. Contrary to the usual two-projective-measurement scheme that is known to destroy quantum features, these fluctuation relations fully capture quantum correlations and quantum coherence at arbitrary times. We further obtain individual integral fluctuation theorems for classical and quantum correlations, as well as for local and global quantum coherences.

A Dynamic Bayesian Network framework for spatial deterioration modelling and reliability updating of timber structures subjected to decay

Reliability assessment of existing timber structures subjected to deterioration processes is an important task to evaluate their serviceability and safety levels. Towards this aim, data collected after inspection campaigns are often used for updating structural reliability and planning future maintenance/inspection activities. Under natural conditions, timber decay involves a large number of uncertainties related to material properties and environmental exposure. These uncertainties are also affected by temporal and spatial variability of associated deterioration processes. In this context, the main objective of this study is to propose a Dynamic Bayesian Network approach for updating the structural reliability of deteriorating timber structures using inspection data. The proposed approach can account for the uncertainties in the decay process and the effect of spatial variability. It is also useful for reliability updating considering the uncertainties of inspection techniques. The proposed methodology is illustrated with the reliability updating of a timber beam subjected to decay deterioration. Results indicate that this approach is useful for evaluating and updating of structural reliability from spatially distributed inspection data. Reliability updating could also be carried out from partial observations at given areas, which is very useful for large-scale infrastructure.

Inference of nonlinear gene regulatory networks through optimized ensemble of support vector regression and dynamic Bayesian networks.

Comprehensive understanding of gene regulatory networks (GRNs) is a major challenge in systems biology. Most methods for modeling and inferring the dynamics of GRNs, such as those based on state space models, vector autoregressive models and G1DBN algorithm, assume linear dependencies among genes. However, this strong assumption does not make for true representation of time-course relationships across the genes, which are inherently nonlinear. Nonlinear modeling methods such as the S-systems and causal structure identification (CSI) have been proposed, but are known to be statistically inefficient and analytically intractable in high dimensions. To overcome these limitations, we propose an optimized ensemble approach based on support vector regression (SVR) and dynamic Bayesian networks (DBNs). The method called SVR-DBN, uses nonlinear kernels of the SVR to infer the temporal relationships among genes within the DBN framework. The two-stage ensemble is further improved by SVR parameter optimization using Particle Swarm Optimization. Results on eight insilico-generated datasets, and two real world datasets of Drosophila Melanogaster and Escherichia Coli, show that our method outperformed the G1DBN algorithm by a total average accuracy of 12%. We further applied our method to model the time-course relationships of ovarian carcinoma. From our results, four hub genes were discovered. Stratified analysis further showed that the expression levels Prostrate differentiation factor and BTG family member 2 genes, were significantly increased by the cisplatin and oxaliplatin platinum drugs; while expression levels of Polo-like kinase and Cyclin B1 genes, were both decreased by the platinum drugs. These hub genes might be potential biomarkers for ovarian carcinoma.

Dynamic Bayesian Networks for Fault Detection, Identification, and Recovery in Autonomous Spacecraft

This paper describes how to exploit the modeling features and inference capabilities of dynamic Bayesian networks (DBN), in designing and implementing an innovative approach to fault detection, identification, and recovery (FDIR) for autonomous spacecrafts (e.g., a Mars rover). In particular, issues like partial observability, uncertain system evolution and system-environment interaction, as well as the prediction and mitigation of imminent failures can be naturally addressed by the proposed approach. The DBN framework can augment the modeling and analytical power of standard FDIR methodologies, while still being able to be integrated into the usual system modeling procedures (like, for instance, fault tree analysis). An FDIR cycle composed of the tasks of diagnosis (identification of the current state of the system), prognosis (identification of the future state under the current conditions), and recovery (selection of the best set of actions the autonomous system can perform, in order to avoid critical situations) is introduced and characterized through a DBN model. In particular, by considering the execution of recovery actions in response to either a current or a future abnormal situation, both reactive as well as preventive recovery can be addressed respectively. The proposed approach has been implemented in an on-board software architecture called Anomaly resolution and prognostic health management for autonomy (ARPHA), realized during the Verifim study funded by the European Space Agency and jointly performed with Thales/Alenia Italy. We report on some of the results obtained by performing a case study concerning the FDIR analysis of the power supply system of the ExoMars rover, by considering different anomalous and failure simulated scenarios; we conclude that ARPHA is able to properly detect and deal with the simulated problems.

Bayesian networks for mathematical models: Techniques for automatic construction and efficient inference

Expert knowledge in the form of mathematical models can be considered sufficient statistics of all prior experimentation in the domain, embodying generic or abstract knowledge of it. When used in a probabilistic framework, such models provide a sound foundation for data mining, inference, and decision making under uncertainty.We describe a methodology for encapsulating knowledge in the form of ordinary differential equations (ODEs) in dynamic Bayesian networks (DBNs). The resulting DBN framework can handle both data and model uncertainty in a principled manner, can be used for temporal data mining with noisy and missing data, and can be used to re-estimate model parameters automatically using data streams. A standard assumption when performing inference in DBNs is that time steps are fixed. Generally, the time step chosen is small enough to capture the dynamics of the most rapidly changing variable. This can result in DBNs having a natural time step that is very short, leading to inefficient inference; this is particularly an issue for DBNs derived from ODEs and for systems where the dynamics are not uniform over time.We propose an alternative to the fixed time step inference used in standard DBNs. In our algorithm, the DBN automatically adapts the time step lengths to suit the dynamics in each step. The resulting system allows us to efficiently infer probable values of hidden variables using multiple time series of evidence, some of which may be sparse, noisy or incomplete.We evaluate our approach with a DBN based on a variant of the van der Pol oscillator, and demonstrate an example where it gives more accurate results than the standard approach, but using only one tenth the number of time steps.We also apply our approach to a real-world example in critical care medicine. By incorporating knowledge in the form of an existing ODE model, we have built a DBN framework for efficiently predicting individualised patient responses using the available bedside and lab data.

Estimating leaf area index from MODIS and surface meteorological data using a dynamic Bayesian network

Remotely sensed data is the main source of vegetation leaf area index (LAI) information on the regional to global scale. Many validation results have revealed that the accuracy of the retrieved LAI is often affected by the cloud cover of imagery, instrument problems, and inversion algorithms. Ground meteorological station data, characterized by relatively high accuracy and time continuity compared with remote sensing data, can provide complementary information to remote sensing observations. In this paper, we combine the potential advantages of both types of data in order to improve LAI retrievals in the Heihe River Basin, an arid and semi-arid area in northwest China where Moderate Resolution Imaging Spectroradiometer (MODIS) LAI values are significantly underestimated. A dynamic Bayesian network (DBN) is used to integrate these two data types for time series LAI estimation. Results show that the square of correlation coefficient between LAI values estimated by our DBN method (referred to as DBN LAI) and field measured LAI values is 0.76, with a root mean square error of 0.78. The DBN LAI are closer to field measurements than the MODIS LAI standard product values. Moreover, by introducing ground meteorological station data using a dynamic process model, DBN LAI show better temporal consistency than the MODIS LAI. It is concluded that the quality of LAI retrievals can be improved by combining remote sensing data and ground meteorological station data using a filtering inference algorithm in a DBN framework. More importantly, the study provides a basis and method for utilizing ground meteorological station network data to estimate land surface parameters on a regional scale.

Driver Alertness Monitoring Using Fusion of Facial Features and Bio-Signals

Driver drowsiness is among the leading causal factors in traffic accidents occurring worldwide. This paper describes a method to monitor driver safety by analyzing information related to fatigue using two distinct methods: eye movement monitoring and bio-signal processing. A monitoring system is designed in Android-based smartphone where it receives sensory data via wireless sensor network and further processes the data to indicate the current driving aptitude of the driver. It is critical that several sensors are integrated and synchronized for a more realistic evaluation of the driver behavior. The sensors applied include a video sensor to capture the driver image and a bio-signal sensor to gather the driver photoplethysmograph signal. A dynamic Bayesian network framework is used for the driver fatigue evaluation. A warning alarm is sounded if driver fatigue is believed to reach a defined threshold. The manifold testing of the system demonstrates the practical use of multiple features, particularly with discrete methods, and their fusion enables a more authentic and ample fatigue detection.

Inferring Nonstationary Gene Networks from Longitudinal Gene Expression Microarrays

Inferring gene networks from longitudinal gene expression microarrays is a crucial step towards the study of gene regulatory mechanisms. A decade ago, expensive microarray technology restricted the number of samples undergoing gene expression profiling in single studies, leading the inference algorithms that assume stationary gene networks to the best solution. Thanks to decreasing cost of modern microarray technologies, more gene expression profiles can be assessed in single studies. With more samples available, we can relax the stationarity assumption and develop a method to infer dynamic gene networks, which can reflect more realistic biology where genes adaptively orchestrate each other. This paper applied the framework of dynamic Bayesian networks to infer adaptive gene interactions by identifying individual transition networks between pairs of consecutive times. Due to high computational burden of inferring the interconnection patterns among all genes over time, we designed a parallelizable inference algorithm to make feasible the task. We validated our approach by two clinical studies: yellow fever vaccination and mechanical periodontal therapy. The inferred dynamic networks achieved more than 90% predictive accuracy, a significant improvement when compared to stationary models (p?<?0.05). The adaptive models can help explain the induction of innate immunology in greater details after yellow fever vaccination and interpret the anti-inflammatory effect of mechanical periodontal therapy.

Discretization of Time Series Data

An increasing number of algorithms for biochemical network inference from experimental data require discrete data as input. For example, dynamic Bayesian network methods and methods that use the framework of finite dynamical systems, such as Boolean networks, all take discrete input. Experimental data, however, are typically continuous and represented by computer floating point numbers. The translation from continuous to discrete data is crucial in preserving the variable dependencies and thus has a significant impact on the performance of the network inference algorithms. We compare the performance of two such algorithms that use discrete data using several different discretization algorithms. One of the inference methods uses a dynamic Bayesian network framework, the other-a time-and state-discrete dynamical system framework. The discretization algorithms are quantile, interval discretization, and a new algorithm introduced in this article, SSD. SSD is especially designed for short time series data and is capable of determining the optimal number of discretization states. The experiments show that both inference methods perform better with SSD than with the other methods. In addition, SSD is demonstrated to preserve the dynamic features of the time series, as well as to be robust to noise in the experimental data. A C++ implementation of SSD is available from the authors at http://polymath.vbi.vt.edu/discretization .

Hand gesture recognition based on dynamic Bayesian network framework

In this paper, we propose a new method for recognizing hand gestures in a continuous video stream using a dynamic Bayesian network or DBN model. The proposed method of DBN-based inference is preceded by steps of skin extraction and modelling, and motion tracking. Then we develop a gesture model for one- or two-hand gestures. They are used to define a cyclic gesture network for modeling continuous gesture stream. We have also developed a DP-based real-time decoding algorithm for continuous gesture recognition. In our experiments with 10 isolated gestures, we obtained a recognition rate upwards of 99.59% with cross validation. In the case of recognizing continuous stream of gestures, it recorded 84% with the precision of 80.77% for the spotted gestures. The proposed DBN-based hand gesture model and the design of a gesture network model are believed to have a strong potential for successful applications to other related problems such as sign language recognition although it is a bit more complicated requiring analysis of hand shapes.

Contributions to the Estimation of Mixed-State Conditionally Heteroscedastic Latent Factor Models: A Comparative Study

Mixed-State conditionally heteroscedastic latent factor models attempt to describe a complex nonlinear dynamic system with a succession of linear latent factor models indexed by a switching variable. Unfortunately, despite the framework's simplicity exact state and parameter estimation are still intractable because of the interdependency across the latent factor volatility processes. Recently, a broad class of learning and inference algorithms for time series models have been successfully cast in the framework of dynamic Bayesian networks (DBN). This paper describes a novel DBN-based switching conditionally heteroscedastic latent factor model. The key methodological contribution of this paper is the novel use of the Generalized Pseudo-Bayesian method GPB2, a structured variational learning approach and an approximated version of the Viterbi algorithm in conjunction with the EM algorithm for overcoming the intractability of exact inference in mixed-state latent factor model. The conditional EM algorithm that we have developed for the maximum likelihood estimation is based on an extended switching Kalman filter approach which yields inferences about the unobservable path of the common factors and their variances, and the latent variable of the state process. Extensive Monte Carlo simulations show promising results for tracking, interpolation, synthesis, and classification using learned models.

Applying dynamic Bayesian networks to perturbed gene expression data.

BackgroundA central goal of molecular biology is to understand the regulatory mechanisms of gene transcription and protein synthesis. Because of their solid basis in statistics, allowing to deal with the stochastic aspects of gene expressions and noisy measurements in a natural way, Bayesian networks appear attractive in the field of inferring gene interactions structure from microarray experiments data. However, the basic formalism has some disadvantages, e.g. it is sometimes hard to distinguish between the origin and the target of an interaction. Two kinds of microarray experiments yield data particularly rich in information regarding the direction of interactions: time series and perturbation experiments. In order to correctly handle them, the basic formalism must be modified. For example, dynamic Bayesian networks (DBN) apply to time series microarray data. To our knowledge the DBN technique has not been applied in the context of perturbation experiments.ResultsWe extend the framework of dynamic Bayesian networks in order to incorporate perturbations. Moreover, an exact algorithm for inferring an optimal network is proposed and a discretization method specialized for time series data from perturbation experiments is introduced. We apply our procedure to realistic simulations data. The results are compared with those obtained by standard DBN learning techniques. Moreover, the advantages of using exact learning algorithm instead of heuristic methods are analyzed.ConclusionWe show that the quality of inferred networks dramatically improves when using data from perturbation experiments. We also conclude that the exact algorithm should be used when it is possible, i.e. when considered set of genes is small enough.

Active Affective State Detection and User Assistance With Dynamic Bayesian Networks

With the rapid development of pervasive and ubiquitous computing applications, intelligent user-assistance systems face challenges of ambiguous, uncertain, and multimodal sensory observations, user's changing state, and various constraints on available resources and costs in making decisions. We introduce a new probabilistic framework based on the dynamic Bayesian networks (DBNs) to dynamically model and recognize user's affective states and to provide the appropriate assistance in order to keep user in a productive state. We incorporate an active sensing mechanism into the DBN framework to perform purposive and sufficing information integration in order to infer user's affective state and to provide correct assistance in a timely and efficient manner. Experiments involving both synthetic and real data demonstrate the feasibility of the proposed framework as well as the effectiveness of the proposed active sensing strategy.

Object-based analysis and interpretation of human motion in sports video sequences by dynamic bayesian networks

In this paper, we present a novel scheme for object-based video analysis and interpretation based on automatic video object extraction, video object abstraction, and semantic event modeling. In our proposed scheme, video objects (VOs) are first segmented, followed by a video object abstraction algorithm for identifying key frames to reduce data redundancy and provide reliable feature data for next stage of the algorithm. Semantic feature modeling scheme is based on temporal variation of low-level features extracted from VOs. More specifically, the Dynamic Bayesian Network (DBN) is used to characterize the spatio-temporal nature of the semantic objects. Comparing to a hidden Markov model (HMM), the DBN framework can provide a detailed description of the characteristics of VOs, which can be readily interfaced to MPEG-7 application. Experimental results that demonstrate the effective performance of the proposed approach are also presented.