Anomaly detection in vessel tracks using Bayesian networks

Real-world entities comprising a complex system evolve as a function of time and respond to external perturbations. Dynamic Bayesian networks extend the fundamental ideas behind static Bayesian networks to model associations arising from the temporal dynamics between the entities of interest. This has to be contrasted with static Bayesian networks, which model the network structure from multiple independent realizations of the entities of a snapshot of the process. More importantly, incorporating the temporal signatures is useful in capturing possible feedback loops that are implicitly disregarded in the case of static Bayesian networks. Since feedback loops are ubiquitous in biological pathways, dynamic Bayesian network modeling is expected to result in better representations of such pathways.

We propose a dynamic differential Bayesian networks (DDBNs) and nonparametric regression model. This model is an extended model of traditional dynamic Bayesian networks (DBNs), which can incorporate temporal information in a natural way and directly handle real-valued data obtained from microarrays without any transformation. In addition, it can cope with differential information between gene expression levels, without any loss to the traditional advantage, i.e., the capability of estimating non-linear relationships between genes. We apply DDBNs to analyze simulated data and real data, i.e., Saccharomyces cerevisiae cell cycle gene expression data. We have confirmed the effectiveness of our approach in the sense that some edges have been successfully detected only by DDBNs, not by DBNs.

Background. Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease primarily affecting upper and lower motor neurons in the brain and spinal cord. The heterogeneity in the course of ALS clinical progression and ultimately survival, coupled with the rarity of this disease, make predicting disease outcome at the level of the individual patient very challenging. Besides, stratification of ALS patients has been known for years as a question of great importance to clinical practice, research and drug development. Methods. In this work, we present a Dynamic Bayesian Network (DBN) model of ALS progression to detect probabilistic relationships among variables included in the Pooled Resource Open-Access ALS Clinical Trials Database (PRO-ACT), which provides records of over 10,700 patients from different clinical trials, and with over 2,869,973 longitudinally collected data measurements. Results. Our model unravels new dependencies among clinical variables in relation to ALS progression, such as the influence of basophil count and creatine kinase on patients’ clinical status and the respiratory functional state, respectively. Furthermore, it provided an indication of ALS temporal evolution, in terms of the most probable disease trajectories across time at the level of both patient population and individual patient. Conclusions. The risk factors identified by out DBN model could allow patients' stratification based on velocity of disease progression and a sensitivity analysis on this latter in response to changes in input variables, i.e. variables measured at diagnosis.

To take full advantage of the historical information, we adopt dynamic Bayesian networks and discrete-time Bayesian networks to describe the event trees or dynamic fault trees. On this basis of the regulation, corresponding dynamic Bayesian networks and discrete-time Bayesian networks models are established for safety assessment in timing system. When the discrete-time time Bayesian network of dynamic fault trees is acquired, the Bayesian network is used to study the consequences probability, key degree and computation method. A simulated banking transaction system is analyzed and assessed by above models. The results show that the safety assessment scheme can provide model without subjectivity, which raises the credibility of model. In addition, it can obtain general assessment results and acquire more extensive diagnostics and information.

Discovering gene relationship from gene expression data is a hot topic in the post-genomic era. In recent years, Bayesian network has become a popular method to reconstruct the gene regulatory network due to the statistical nature. However, it is not suitable for analyzing the time-series data and cannot deal with cycles in the gene regulatory network. In this paper we apply the dynamic Bayesian network to model the gene relationship in order to overcome these difficulties. By incorporating the structural expectation maximization algorithm into the dynamic Bayesian network model, we develop a new method to learn the regulatory network from the S.Cerevisiae cell cycle gene expression data. The experimental results demonstrate that the accuracy of our method outperforms the previous work

Tracing and prediction analysis of an urban pipeline leakage accident based on the catastrophe DBN model

This paper proposes a method to localize a moving agent - such as an autonomous surveillance vehicle - inside a known environment using First Person Viewpoint video data. Anomalies w.r.t. expected vehicle motion and image content are extracted to guide the localization, signal when the localization results are not trustworthy, and explain the reason for the failure. During the training phase, a Dynamic Bayesian Network model is learned, which couples positional and video data. To learn it, clustering is performed on the odometry data, and a modified Kalman Variational Autoencoder is built over the video data. During the testing phase, a Coupled Markov Jump Particle Filter leverages the learned Dynamic Bayesian Network to extract anomalies and to estimate the vehicle’s position, given only camera data. The proposed method is evaluated on two real-world datasets of a vehicle performing perimeter monitoring of a closed environment [17] and of a shopping cart moving in a supermarket

The cell functions and development are regulated by complex networks of genes, proteins and other components by means of their mutual interactions. These networks are called gene regulatory networks (GRNs). GRNs are used to reveal the fundamental gene regulatory mechanisms, to determine the reasons for many diseases and interactions between drugs and their targets. The introduction of experimental technologies such as microarrays, ChIP-chip which combines chromatin immunoprecipitation (ChIP) with microarrays and ChIP-Seq which combines ChIP with DNA sequencing, has provided a large number of available datasets related to gene expression and transcription factors (TFs) and their interactions. These datasets are basis for further analysis to reveal the gene regulation mechanisms. Many models have been applied to represent gene regulatory networks. We have used the dynamic Bayesian network model which is able to cope with missing data and can include a prior knowledge about transcription factors and their activation/inhibition of corresponding genes. We describe the obtained results and survey the common structure learning algorithms for learning of GRN's structure. We tested the obtained GRN for test datasets with different sizes and in the paper describe obtained dependencies between the ratio of Bayesian score and BIC and dataset size.

Illegal fishing is a global problem affecting society in numerous ways: from impacting the sustainability of artisanal and subsistence fishing communities to disturbing the balance of delicate oceanic ecosystems through the systematic overexploitation of fish stocks. On a global level, the success criminal organizations have had with hiding their activities is largely attributed to the practice of transshipment (where a fishing vessel offloads its catch to a transport vessel at sea), which makes tracing the origins of illegally caught fish extremely difficult. Extensive research has been conducted to leverage data from the Automatic Identification System (AIS) to identify vessels fishing illegally, but current methods often overlook interactions between vessels and have not been shown to affect the global extent of illegal fishing. In this paper, transshipment encounters are modelled as a social network graph where nodes represent vessels and edges represent transshipment encounters between them. A static analysis of the transshipment networks is conducted to identify key vessels enabling illegal fishing through transshipment. A dynamic visualization tool is then implemented to provide context on the formation and evolution of transshipment networks over time. Based on the dynamic analysis of the network, insights are gained about different ways in which transshipment encounters might be used to offload marine products. A dynamic analysis of criminal subnetworks is conducted, showcasing how network modelling can be used to enable a global approach in the combat against illegal fishing.

In this article, we develop and evaluate three Dynamic Bayesian Network (DBN) models for assessing temporally variable learner scientific inquiry skills (Hypothesis Generation and Variable Identification) in INQPRO learning environment. Empirical studies were carried out to examine the matching accuracies and identify the models’ drawbacks. We demonstrate how the insights gained from a preceding model have eventually led to the improvement of subsequent models. In this study, the entire evaluation process involved 6 domain experts and 61 human learners. The matching accuracies of the models are measured by (1) comparing with the results gathered from the pretest, posttest, and learner’s self-rating scores; and (2) comments given by domain experts based on learners’ interaction logs and the graph patterns exhibited by the models.KeywordsScientific Inquiry SkillsDynamic Bayesian Networks

Fire probability prediction of offshore platform based on Dynamic Bayesian Network

Bayesian network models provide an attractive framework for multimodal sensor fusion. They combine an intuitive graphical representation with efficient algorithms for inference and learning. However, the unsupervised nature of standard parameter learning algorithms for Bayesian networks can lead to poor performance in classification tasks. We have developed a supervised learning framework for Bayesian networks, which is based on the Adaboost algorithm of Schapire and Freund. Our framework covers static and dynamic Bayesian networks with both discrete and continuous states. We have tested our framework in the context of a novel multimodal HCI application: a speech-based command and control interface for a Smart Kiosk. We provide experimental evidence for the utility of our boosted learning approach.

A probabilistic framework based on the Bayesian networks for modeling and real-time inferring human fatigue by integrating information from various sensory data and certain relevant contextual information is introduced. A static fatigue model that captures the static relationships between fatigue, significant factors that cause fatigue, and various sensory observations that typically result from fatigue is first presented. Such a model provides mathematically coherent and sound basis for systematically aggregating uncertain evidences from different sources, augmented with relevant contextual information. The static model, however, fails to capture the dynamic aspect of fatigue. Fatigue is a cognitive state that is developed over time. To account for the temporal aspect of human fatigue, the static fatigue model is extended based on dynamic Bayesian networks. The dynamic fatigue model allows to integrate fatigue evidences not only spatially but also temporally, therefore, leading to a more robust and accurate fatigue modeling and inference. A real-time nonintrusive fatigue monitor was built based on integrating the proposed fatigue model with a computer vision system developed for extracting various visual cues typically related to fatigue. Performance evaluation of the fatigue monitor using both synthetic and real data demonstrates the validity of the proposed fatigue model in both modeling and real-time inference of fatigue.

Modelling based on probabilistic inference can be used to estimate the quality of information delivered by a military sensor network. Different modelling tools have complementary characteristics that can be leveraged to create an accurate model open to intuitive and efficient querying. In particular, stochastic process models can be used to abstract away from the physical reality by describing it as components that exist in discrete states with probabilistically invoked actions that change the state. The quality of information may be assessed by using the model to compute the probability that reports made by the network to its users are correct. In contrast, dynamic Bayesian network models, which have been used in a variety of military applications, are a more suitable vehicle for understanding the overall network performance and making inferences about the quality of information. Queries can be made simply by instantiating some variables and computing the probability distributions over others. We show that it is possible to combine both modelling tools by constructing a Bayesian network over the state variables of the process algebra model. The sparsity of the resulting Bayesian network allows fast propagation of probabilities, and hence interactive querying for the quality of information.

Driver distraction has become an important and growing safety concern as the use of in-vehicle information systems (IVISs), such as cell phones and navigation systems, continues to increase. One approach to allowing people to benefit from IVISs without compromising safety is to create adaptive IVISs that adjust their functions according to driver and roadway state. A critical element of adaptive IVISs involves monitoring driver distraction in real time; with such a monitoring function it is possible to mitigate that distraction. This study applied Bayesian networks (BNs), a data mining method, to develop a real-time approach to detecting cognitive distraction using drivers’ eye movements and driving performance. Data were collected in a simulator experiment involving 10 participants who interacted with an IVIS system while driving. BN models were trained and tested to investigate the influence of three model characteristics on distraction detection: time history of driver behavior, inclusion of hidden nodes in the model structure, and how data are summarized and the length of training sequences. Results showed that BNs could identify driver distraction reliably with an average accuracy of 80.1%. Dynamic BNs (DBNs) that consider time dependencies of driver behavior produced more sensitive models than static BNs (SBNs). Longer training sequences improved DBN model performance. Blink frequency and fixation measures were particularly indicative of distraction. These results demonstrate that BNs, especially DBNs, can detect driver cognitive distraction by extracting information from complex behavioral data. Potential applications include the design of adaptive in-vehicle systems and the evaluation of driver distraction.