Strategically Patrolling in a Chemical Cluster Addressing Gas Pollutants' Releases through a Game-Theoretic Model.

Chemical production activities in chemical clusters, if not well managed, will pose great threats to the surrounding air environment and impose great burden on emergency handling. Therefore, it is urgent and substantial in a chemical cluster to develop proper and suitable pollution controlling strategies for an inspection agency to monitor chemical production processes. Apart from the static monitoring resources (e.g., monitoring stations and gas sensor modules), patrolling by mobile vehicle resources is arranged for better detecting the illegal releasing behaviors of emission spots in different chemical plants. However, it has been proven that the commonly used patrolling strategies (i.e., the fixed route strategy and the purely randomized route strategy) are non-optimal and fail to interact with intelligent chemical plants. Therefore, we proposed the Chemical Cluster Environmental Protection Patrolling (CCEPP) game to tackle the problem in this paper. Through combining the source estimation process, the game is modeled to detect the illegal releasing behaviors of chemical plants by randomly and strategically arranging the patrolling routes and intensities in different chemical sites. In this game-theoretic model, players (patroller and chemical sites), strategies, payoffs, and game solvers are modeled in sequence. More importantly, this game model also considers traffic delays or bounded cognition of patrollers on patrolling plans. Therefore, a discrete Markov decision process was used to model this stochastic process. Further, the model is illustrated by a case study. Results imply that the patrolling strategy suggested by the CCEPP game outperforms both the fixed route strategy and the purely randomized route strategy.

Evolution Model of Spatial Interaction Network in Online Social Networking Services.

The development of online social networking services provides a rich source of data of social networks including geospatial information. More and more research has shown that geographical space is an important factor in the interactions of users in social networks. In this paper, we construct the spatial interaction network from the city level, which is called the city interaction network, and study the evolution mechanism of the city interaction network formed in the process of information dissemination in social networks. A network evolution model for interactions among cities is established. The evolution model consists of two core processes: the edge arrival and the preferential attachment of the edge. The edge arrival model arranges the arrival time of each edge; the model of preferential attachment of the edge determines the source node and the target node of each arriving edge. Six preferential attachment models (Random-Random, Random-Degree, Degree-Random, Geographical distance, Degree-Degree, Degree-Degree-Geographical distance) are built, and the maximum likelihood approach is used to do the comparison. We find that the degree of the node and the geographic distance of the edge are the key factors affecting the evolution of the city interaction network. Finally, the evolution experiments using the optimal model DDG are conducted, and the experiment results are compared with the real city interaction network extracted from the information dissemination data of the WeChat web page. The results indicate that the model can not only capture the attributes of the real city interaction network, but also reflect the actual characteristics of the interactions among cities.

Data-Driven Hazardous Gas Dispersion Modeling Using the Integration of Particle Filtering and Error Propagation Detection.

The accurate prediction of hazardous gas dispersion process is essential to air quality monitoring and the emergency management of contaminant gas leakage incidents in a chemical cluster. Conventional Gaussian-based dispersion models can seldom give accurate predictions due to inaccurate input parameters and the computational errors. In order to improve the prediction accuracy of a dispersion model, a data-driven air dispersion modeling method based on data assimilation is proposed by applying particle filter to Gaussian-based dispersion model. The core of the method is continually updating dispersion coefficients by assimilating observed data into the model during the calculation process. Another contribution of this paper is that error propagation detection rules are proposed to evaluate their effects since the measured and computational errors are inevitable. So environmental protection authorities can be informed to what extent the model output is of high confidence. To test the feasibility of our method, a numerical experiment utilizing the SF6 concentration data sampled from an Indianapolis field study is conducted. Results of accuracy analysis and error inspection imply that Gaussian dispersion models based on particle filtering and error propagation detection have better performance than traditional dispersion models in practice though sacrificing some computational efficiency.

Comparison of Machine Learning Models for Hazardous Gas Dispersion Prediction in Field Cases.

Dispersion prediction plays a significant role in the management and emergency response to hazardous gas emissions and accidental leaks. Compared with conventional atmospheric dispersion models, machine leaning (ML) models have both high accuracy and efficiency in terms of prediction, especially in field cases. However, selection of model type and the inputs of the ML model are still essential problems. To address this issue, two ML models (i.e., the back propagation (BP) network and support vector regression (SVR) with different input selections (i.e., original monitoring parameters and integrated Gaussian parameters) are proposed in this paper. To compare the performances of presented ML models in field cases, these models are evaluated using the Prairie Grass and Indianapolis field data sets. The influence of the training set scale on the performances of ML models is analyzed as well. Results demonstrate that the integrated Gaussian parameters indeed improve the prediction accuracy in the Prairie Grass case. However, they do not make much difference in the Indianapolis case due to their inadaptability to the complex terrain conditions. In addition, it can be summarized that the SVR shows better generalization ability with relatively small training sets, but tends to under-fit the training data. In contrast, the BP network has a stronger fitting ability, but sometimes suffers from an over-fitting problem. As a result, the model and input selection presented in this paper will be of great help to environmental and public health protection in real applications.

An Extended Chemical Plant Environmental Protection Game on Addressing Uncertainties of Human Adversaries.

Chemical production activities in industrial districts pose great threats to the surrounding atmospheric environment and human health. Therefore, developing appropriate and intelligent pollution controlling strategies for the management team to monitor chemical production processes is significantly essential in a chemical industrial district. The literature shows that playing a chemical plant environmental protection (CPEP) game can force the chemical plants to be more compliant with environmental protection authorities and reduce the potential risks of hazardous gas dispersion accidents. However, results of the current literature strictly rely on several perfect assumptions which rarely hold in real-world domains, especially when dealing with human adversaries. To address bounded rationality and limited observability in human cognition, the CPEP game is extended to generate robust schedules of inspection resources for inspection agencies. The present paper is innovative on the following contributions: (i) The CPEP model is extended by taking observation frequency and observation cost of adversaries into account, and thus better reflects the industrial reality; (ii) Uncertainties such as attackers with bounded rationality, attackers with limited observation and incomplete information (i.e., the attacker's parameters) are integrated into the extended CPEP model; (iii) Learning curve theory is employed to determine the attacker's observability in the game solver. Results in the case study imply that this work improves the decision-making process for environmental protection authorities in practical fields by bringing more rewards to the inspection agencies and by acquiring more compliance from chemical plants.

A data-driven approach for optimal design of integrated air quality monitoring network in a chemical cluster.

The chemical industry is of paramount importance to the world economy and this industrial sector represents a substantial income source for developing countries. However, the chemical plants producing inside an industrial district pose a great threat to the surrounding atmospheric environment and human health. Therefore, designing an appropriate and available air quality monitoring network (AQMN) is essential for assessing the effectiveness of deployed pollution-controlling strategies and facilities. As monitoring facilities located at inappropriate sites would affect data validity, a two-stage data-driven approach constituted of a spatio-temporal technique (i.e. Bayesian maximum entropy) and a multi-objective optimization model (i.e. maximum concentration detection capability and maximum dosage detection capability) is proposed in this paper. The approach aims at optimizing the design of an AQMN formed by gas sensor modules. Owing to the lack of long-term measurement data, our developed atmospheric dispersion simulation system was employed to generate simulated data for the above method. Finally, an illustrative case study was implemented to illustrate the feasibility of the proposed approach, and results imply that this work is able to design an appropriate AQMN with acceptable accuracy and efficiency.

Playing Chemical Plant Environmental Protection Games with Historical Monitoring Data.

The chemical industry is very important for the world economy and this industrial sector represents a substantial income source for developing countries. However, existing regulations on controlling atmospheric pollutants, and the enforcement of these regulations, often are insufficient in such countries. As a result, the deterioration of surrounding ecosystems and a quality decrease of the atmospheric environment can be observed. Previous works in this domain fail to generate executable and pragmatic solutions for inspection agencies due to practical challenges. In addressing these challenges, we introduce a so-called Chemical Plant Environment Protection Game (CPEP) to generate reasonable schedules of high-accuracy air quality monitoring stations (i.e., daily management plans) for inspection agencies. First, so-called Stackelberg Security Games (SSGs) in conjunction with source estimation methods are applied into this research. Second, high-accuracy air quality monitoring stations as well as gas sensor modules are modeled in the CPEP game. Third, simplified data analysis on the regularly discharging of chemical plants is utilized to construct the CPEP game. Finally, an illustrative case study is used to investigate the effectiveness of the CPEP game, and a realistic case study is conducted to illustrate how the models and algorithms being proposed in this paper, work in daily practice. Results show that playing a CPEP game can reduce operational costs of high-accuracy air quality monitoring stations. Moreover, evidence suggests that playing the game leads to more compliance from the chemical plants towards the inspection agencies. Therefore, the CPEP game is able to assist the environmental protection authorities in daily management work and reduce the potential risks of gaseous pollutants dispersion incidents.

An Opinion Interactive Model Based on Individual Persuasiveness.

In order to study the formation process of group opinion in real life, we put forward a new opinion interactive model based on Deffuant model and its improved models in this paper because current models of opinion dynamics lack considering individual persuasiveness. Our model has following advantages: firstly persuasiveness is added to individual's attributes reflecting the importance of persuasiveness, which means that all the individuals are different from others; secondly probability is introduced in the course of interaction which simulates the uncertainty of interaction. In Monte Carlo simulation experiments, sensitivity analysis including the influence of randomness, initial persuasiveness distribution, and number of individuals is studied at first; what comes next is that the range of common opinion based on the initial persuasiveness distribution can be predicted. Simulation experiment results show that when the initial values of agents are fixed, no matter how many times independently replicated experiments, the common opinion will converge at a certain point; however the number of iterations will not always be the same; the range of common opinion can be predicted when initial distribution of opinion and persuasiveness are given. As a result, this model can reflect and interpret some phenomena of opinion interaction in realistic society.

The Large Scale Machine Learning in an Artificial Society: Prediction of the Ebola Outbreak in Beijing.

Ebola virus disease (EVD) distinguishes its feature as high infectivity and mortality. Thus, it is urgent for governments to draw up emergency plans against Ebola. However, it is hard to predict the possible epidemic situations in practice. Luckily, in recent years, computational experiments based on artificial society appeared, providing a new approach to study the propagation of EVD and analyze the corresponding interventions. Therefore, the rationality of artificial society is the key to the accuracy and reliability of experiment results. Individuals' behaviors along with travel mode directly affect the propagation among individuals. Firstly, artificial Beijing is reconstructed based on geodemographics and machine learning is involved to optimize individuals' behaviors. Meanwhile, Ebola course model and propagation model are built, according to the parameters in West Africa. Subsequently, propagation mechanism of EVD is analyzed, epidemic scenario is predicted, and corresponding interventions are presented. Finally, by simulating the emergency responses of Chinese government, the conclusion is finally drawn that Ebola is impossible to outbreak in large scale in the city of Beijing.

Semi-Supervised Projective Non-Negative Matrix Factorization for Cancer Classification.

Advances in DNA microarray technologies have made gene expression profiles a significant candidate in identifying different types of cancers. Traditional learning-based cancer identification methods utilize labeled samples to train a classifier, but they are inconvenient for practical application because labels are quite expensive in the clinical cancer research community. This paper proposes a semi-supervised projective non-negative matrix factorization method (Semi-PNMF) to learn an effective classifier from both labeled and unlabeled samples, thus boosting subsequent cancer classification performance. In particular, Semi-PNMF jointly learns a non-negative subspace from concatenated labeled and unlabeled samples and indicates classes by the positions of the maximum entries of their coefficients. Because Semi-PNMF incorporates statistical information from the large volume of unlabeled samples in the learned subspace, it can learn more representative subspaces and boost classification performance. We developed a multiplicative update rule (MUR) to optimize Semi-PNMF and proved its convergence. The experimental results of cancer classification for two multiclass cancer gene expression profile datasets show that Semi-PNMF outperforms the representative methods.

Online multi-modal robust non-negative dictionary learning for visual tracking.

Dictionary learning is a method of acquiring a collection of atoms for subsequent signal representation. Due to its excellent representation ability, dictionary learning has been widely applied in multimedia and computer vision. However, conventional dictionary learning algorithms fail to deal with multi-modal datasets. In this paper, we propose an online multi-modal robust non-negative dictionary learning (OMRNDL) algorithm to overcome this deficiency. Notably, OMRNDL casts visual tracking as a dictionary learning problem under the particle filter framework and captures the intrinsic knowledge about the target from multiple visual modalities, e.g., pixel intensity and texture information. To this end, OMRNDL adaptively learns an individual dictionary, i.e., template, for each modality from available frames, and then represents new particles over all the learned dictionaries by minimizing the fitting loss of data based on M-estimation. The resultant representation coefficient can be viewed as the common semantic representation of particles across multiple modalities, and can be utilized to track the target. OMRNDL incrementally learns the dictionary and the coefficient of each particle by using multiplicative update rules to respectively guarantee their non-negativity constraints. Experimental results on a popular challenging video benchmark validate the effectiveness of OMRNDL for visual tracking in both quantity and quality.

An effective immunization strategy for airborne epidemics in modular and hierarchical social contact network.

Social contact between individuals is the chief factor for airborne epidemic transmission among the crowd. Social contact networks, which describe the contact relationships among individuals, always exhibit overlapping qualities of communities, hierarchical structure and spatial-correlated. We find that traditional global targeted immunization strategy would lose its superiority in controlling the epidemic propagation in the social contact networks with modular and hierarchical structure. Therefore, we propose a hierarchical targeted immunization strategy to settle this problem. In this novel strategy, importance of the hierarchical structure is considered. Transmission control experiments of influenza H1N1 are carried out based on a modular and hierarchical network model. Results obtained indicate that hierarchical structure of the network is more critical than the degrees of the immunized targets and the modular network layer is the most important for the epidemic propagation control. Finally, the efficacy and stability of this novel immunization strategy have been validated as well.

Mathematical and computational approaches to epidemic modeling: a comprehensive review.

Mathematical and computational approaches are important tools for understanding epidemic spread patterns and evaluating policies of disease control. In recent years, epidemiology has become increasingly integrated with mathematics, sociology, management science, complexity science, and computer science. The cross of multiple disciplines has caused rapid development of mathematical and computational approaches to epidemic modeling. In this article, we carry out a comprehensive review of epidemic models to provide an insight into the literature of epidemic modeling and simulation. We introduce major epidemic models in three directions, including mathematical models, complex network models, and agent-based models. We discuss the principles, applications, advantages, and limitations of these models. Meanwhile, we also propose some future research directions in epidemic modeling.

Accuracy of lattice Monte Carlo simulation of biased diffusion models.

The accuracy of lattice Monte Carlo (LMC) simulation of biased diffusion models is of great importance as far as the simulation credibility is concerned. It is known that the fixed time step LMC algorithm can reproduce the mean and the variance of the particle displacement exactly for all discrete time steps. Thereby, we propose to use the skewness and other quantities to measure the accuracy. For the one-dimensional fixed time step LMC simulation, we obtain an explicit expression for the skewness and find that the algorithm always produces a negative skewness that converges to zero in the long-time limit when the velocity is positive. It is proved that the skewness is inversely proportional to the square root of the number of simulation steps and the first step error only depends on the Péclet number. We further discuss several other measures of the accuracy of the approximation based on appropriately defined mean-square errors, leading to interesting, unexpected results. The accuracy measures can exhibit complicated nonmonotonic behavior and the optimal step size may depend on the measure of accuracy used.