Toxic gas dispersion prediction for point source emission using deep learning method

Abstract

Accurate and rapid toxic gas concentration prediction model plays an important role in emergency aid of sudden gas leak. However, it is difficult for existing dispersion model to achieve accuracy and efficiency requirements at the same time. Although some researchers have considered developing new forecasting models with traditional machine learning, such as back propagation (BP) neural network, support vector machine (SVM), the prediction results obtained from such models need to be improved still in terms of accuracy. Then new prediction models based on deep learning are proposed in this paper. Deep learning has obvious advantages over traditional machine learning in prediction and classification. Deep belief networks (DBNs) as well as convolution neural networks (CNNs) are used to build new dispersion models here. Both models are compared with Gaussian plume model, computation fluid dynamics (CFD) model and models based on traditional machine learning in terms of accuracy, prediction time, and computation time. The experimental results turn out that CNNs model performs better considering all evaluation indexes.     

PSDS 2.0: 一个基于GIS和多个模型的生物潜在分布地预测系统

根据对生物分布地预测模型和软件发展现状的分析和总结, 本研究在PSDS 1.0的基础上提出并实现一个基于GIS且具有多个代表性模型的生物分布地预测系统(PSDS 2.0)。PSDS 2.0系统继承了1.0的环境包络和聚类包络模型, 进一步引入了限制因子包络、马氏距离、支持向量机等新模型, 并针对本领域中模型比较与选择的难点增加了迭代交叉验证的多模型选择功能。系统还实现了灵活定制和评估伪负样本的功能, 通过用只需要正样本的I类模型预测的结果对随机产生的伪负样本进行评估, 减小其落入适宜地区的概率, 进一步提高需要正负样本的II类模型的准确率。GIS功能在PSDS 2.0中也得到加强, 被应用于数据准备及结果分析等重要环节。文章最后以白冠长尾雉(Syrmaticus reevesii)为例, 运用PSDS 2.0系统预测其在中国范围内的潜在分布地, 并对各种模型的预测结果进行评估和比较。     

Two‐stage orthogonality based estimation for semiparametric varying‐coefficient models and its applications in analyzing AIDS data

Semiparametric smoothing methods are usually used to model longitudinal data, and the interest is to improve efficiency for regression coefficients. This paper is concerned with the estimation in semiparametric varying‐coefficient models (SVCMs) for longitudinal data. By the orthogonal projection method, local linear technique, quasi‐score estimation, and quasi‐maximum likelihood estimation, we propose a two‐stage orthogonality‐based method to estimate parameter vector, coefficient function vector, and covariance function. The developed procedures can be implemented separately and the resulting estimators do not affect each other. Under some mild conditions, asymptotic properties of the resulting estimators are established explicitly. In particular, the asymptotic behavior of the estimator of coefficient function vector at the boundaries is examined. Further, the finite sample performance of the proposed procedures is assessed by Monte Carlo simulation experiments. Finally, the proposed methodology is illustrated with an analysis of an acquired immune deficiency syndrome (AIDS) dataset.     

Path-finding in real and simulated rats: assessing the influence of path characteristics on navigation learning

A large body of experimental evidence suggests that the hippocampal place field system is involved in reward based navigation learning in rodents. Reinforcement learning (RL) mechanisms have been used to model this, associating the state space in an RL-algorithm to the place-field map in a rat. The convergence properties of RL-algorithms are affected by the exploration patterns of the learner. Therefore, we first analyzed the path characteristics of freely exploring rats in a test arena. We found that straight path segments with mean length 23 cm up to a maximal length of 80 cm take up a significant proportion of the total paths. Thus, rat paths are biased as compared to random exploration. Next we designed a RL system that reproduces these specific path characteristics. Our model arena is covered by overlapping, probabilistically firing place fields (PF) of realistic size and coverage. Because convergence of RL-algorithms is also influenced by the state space characteristics, different PF-sizes and densities, leading to a different degree of overlap, were also investigated. The model rat learns finding a reward opposite to its starting point. We observed that the combination of biased straight exploration, overlapping coverage and probabilistic firing will strongly impair the convergence of learning. When the degree of randomness in the exploration is increased, convergence improves, but the distribution of straight path segments becomes unrealistic and paths become 'wiggly'. To mend this situation without affecting the path characteristic two additional mechanisms are implemented: a gradual drop of the learned weights (weight decay) and path length limitation, which prevents learning if the reward is not found after some expected time. Both mechanisms limit the memory of the system and thereby counteract effects of getting trapped on a wrong path. When using these strategies individually divergent cases get substantially reduced and for some parameter settings no divergence was found anymore at all. Using weight decay and path length limitation at the same time, convergence is not much improved but instead time to convergence increases as the memory limiting effect is getting too strong. The degree of improvement relies also on the size and degree of overlap (coverage density) in the place field system. The used combination of these two parameters leads to a trade-off between convergence and speed to convergence. Thus, this study suggests that the role of the PF-system in navigation learning cannot be considered independently from the animals' exploration pattern.     

Improvement, Comparison, and Application of Field Measurement Methods for Grassland Vegetation Fractional Coverage

As one of the important vegetation parameters, vegetation fractional coverage （VFC） is more difficult to measure accurately among a good many parameters of plant communities. The temperate typical steppe in the north of China was chosen for investigation in the present study and a digital camera was used to measure herb community coverage in the field, adopting methods of ocular estimation, gridding measurement, visual interpretation, supervised classification, and information extraction of color spatial transformation to calculate the VFC of images captured by the digital camera. In addition VFC calculated by various methods was analyzed and compared VFC, enabling us to propose an effective method for measuring VFC using a digital camera. The results of the present study indicate that： （i） as two common useful and effective methods of measuring VFC with a digital camera, not only does the error of estimated values of visual estimation and supervised classification vary considerably, but the degree of automatization is very low and depends, to a great extent, on the manipulator; （ii） although the method of visual interpretation may assure the precision of the calculated VFC and enable the precision of results obtained using other methods to be determined, as far as large quantities of data are concerned, this method has the disadvantages of wasting time and energy, and the applications of this method are limited; （iii） the precision and stability of VFC calculated using the grid and node method are superior to those of visual estimation and supervised classification and inferior to those of visual interpretation, but, as for visual interpretation and supervised classification, gridding measurements are difficult to apply in practice because they are not time efficient; and （iv） in terms of the precision of calculation of the VFC, an information-extracting model based on an intensity, hue, saturation （IHS） color space-multi-component series segmentation strategy is superior to methods of ocular estimation, gridding measurement, and supervised classification. In terms of practical efficiency, the information-extracting model is superior to visual interpretation, supervised classification, and gridding measurement. It has been proven that estimating the VFC of the north temperate typical steppe using this model is feasible. This is very fundamental research work in grassland ecology.     

Spatial mixture multiscale modeling for aggregated health data

One of the main goals in spatial epidemiology is to study the geographical pattern of disease risks. For such purpose, the convolution model composed of correlated and uncorrelated components is often used. However, one of the two components could be predominant in some regions. To investigate the predominance of the correlated or uncorrelated component for multiple scale data, we propose four different spatial mixture multiscale models by mixing spatially varying probability weights of correlated (CH) and uncorrelated heterogeneities (UH). The first model assumes that there is no linkage between the different scales and, hence, we consider independent mixture convolution models at each scale. The second model introduces linkage between finer and coarser scales via a shared uncorrelated component of the mixture convolution model. The third model is similar to the second model but the linkage between the scales is introduced through the correlated component. Finally, the fourth model accommodates for a scale effect by sharing both CH and UH simultaneously. We applied these models to real and simulated data, and found that the fourth model is the best model followed by the second model.     

基于遥感图像不同辐射校正水平的植被覆盖度估算模型

选用南京市SPOT 5 HRG图像的地物反射率(PAC)、表观反射率(TOA)和灰度值（DN）影像,提取了4种植被指数（VI）,即归一化植被指数（NDVI）、转换植被指数（TVI）、土壤调节植被指数（SAVI）和修正的土壤调节植被指数（MSAVI）,与地面实测的植被覆盖度进行了回归分析,并建立了36个VI-VFC关系模型.结果表明：在所有模型中,基于PAC级影像提取的NDVI和TVI的3次多项式模型最优;其次为基于DN级影像提取的SAVI和MSAVI的3次多项式模型,在VFC>0.8时其精度略高于前两种模型.这4个模型在植被中等密集区域（VFC＝0.4～0.8）的精度高于植被稀疏区域（VFC=0～0.4）.所建模型可通过中间模型的联结,进行推广使用.在基于VI-VFC关系建模过程中,基于遥感影像不同辐射校正水平提取植被指数,有利于充分挖掘遥感影像信息,进而提高VFC估算的精度.     

Shoulder Kinematics and Spatial Pattern of Trapezius Electromyographic Activity in Real and Virtual Environments

The design of an industrial workstation tends to include ergonomic assessment steps based on a digital mock-up and a virtual reality setup. Lack of interaction and system fidelity is often reported as a main issue in such virtual reality applications. This limitation is a crucial issue as thorough ergonomic analysis is required for an investigation of the biomechanics. In the current study, we investigated the biomechanical responses of the shoulder joint in a simulated assembly task for comparison with the biomechanical responses in virtual environments. Sixteen male healthy novice subjects performed the task on three different platforms: real (RE), virtual (VE), and virtual environment with force feedback (VEF) with low and high precision demands. The subjects repeated the task 12 times (i.e., 12 cycles). High density electromyography from the upper trapezius and rotation angles of the shoulder joint were recorded and split into the cycles. The angular trajectories and velocity profiles of the shoulder joint angles over a cycle were computed in 3D. The inter-subject similarity in terms of normalized mutual information on kinematics and electromyography was investigated. Compared with RE the task in VE and VEF was characterized by lower kinematic maxima. The inter-subject similarity in RE compared with intra-subject similarity across the platforms was lower in terms of movement trajectories and greater in terms of trapezius muscle activation. The precision demand resulted in lower inter- and intra-subject similarity across platforms. The proposed approach identifies biomechanical differences in the shoulder joint in both VE and VEF compared with the RE platform, but these differences are less marked in VE mostly due to technical limitations of co-localizing the force feedback system in the VEF platform.     

基于动态方向梯度矢量流模型的脑肿瘤图像分割

目的:针对GVF Snake模型算法收敛容易陷入局部极小值及对初始轮廓位置敏感等缺点,提出一种动态方向梯度矢量流模型(DDGVF),使其更适合医学图像的分割。方法:利用主动轮廓模型的提取和跟踪特定区域内目标轮廓的方法,将其应用于医学图像如CT、MRI和超声图像的处理,以获取特定器官及组织的轮廓。结果:动态方向梯度矢量流场(DDGVF)能够较好地提取出脑肿瘤图像。结论:利用该方法能够较好地分割提取出脑肿瘤图像的肿瘤病变区域,为进一步对其纹理和形状等特征进行描述和分析提供了可靠的依据。     

Finite element implementation of a generalized Fung-elastic constitutive model for planar soft tissues

Numerical simulations of the anisotropic mechanical properties of soft tissues and tissue-derived biomaterials using accurate constitutive models remain an important and challenging research area in biomechanics. While most constitutive modeling efforts have focused on the characterization of experimental data, only limited studies are available on the feasibility of utilizing those models in complex computational applications. An example is the widely utilized exponential constitutive model proposed by Fung. Although present in the biomechanics literature for several decades, implementation of this model into finite element (FE) simulations has been limited. A major reason for limited numerical implementations are problems associated with inherent numerical instability and convergence. To address this issue, we developed and applied two restrictions for a generalized Fung-elastic constitutive model necessary to achieve numerical stability. These are (1) convexity of the strain energy function, and (2) the condition number of material stiffness matrix set lower than a prescribed value. These constraints were implemented in the nonlinear regression used for constitutive model parameter estimation to the experimental biaxial mechanical data. We then implemented the generalized Fung-elastic model into a commercial FE code (ABAQUS, Pawtucket, RI, USA). Single element and multi-element planar biaxial test simulations were conducted to verify the accuracy and robustness of the implementation. Results indicated that numerical convergence and accurate FE implementation were consistently obtained. The present study thus presents an integrated framework for accurate and robust implementation of pseudo-elastic constitutive models for planar soft tissues. Moreover, since our approach is formulated within a general FE code, it can be straightforwardly adopted across multiple software platforms.     

Influence of Electric Field Operation Modes on Nickel Migration during Electrokinetic Treatment

Nickel-contaminated sediment can be remediated by electrokinetic techniques. In practical application, nickel may migrate into deep layers and potentially pollute unpolluted layers. In this study, the influence of different types of electric field operation modes on nickel migration in sediment layers was studied. The following experiments were conducted: without current application (NC), with horizontal electric field (HEF), with vertical electric field (VEF), and 2D electric field (horizontal and vertical). The following Ni overall removal efficacies were achieved: 48.9, 49.4, 55.7, and 58.1 after VEF, NC, 2D, and HEF, respectively. Using a vertical electric field in 2D and VEF resulted in less Ni found in the vertical sediment bed (25% and 60%, respectively). The available Ni content in a contaminated sediment bed decreased significantly after using the electrokinetic treatments: 46%, 62%, and 79% after HEF, VEF, and 2D, respectively. Vertical electric field is the most efficient for Ni removal in terms of the amount of Ni that migrated into deeper layers of sediment bed and its amount in mobile form. Thus, the vertical electric field is a promising and practical method for the remediation of nickel-contaminated sediments. Risk analysis indicated the efficiency of EK treatment as available Ni content was reduced after the applied treatments to a degree that poses no risk for the environment.     

Computational models of visual neurons specialised in the detection of periodic and aperiodic oriented visual stimuli: bar and grating cells

Computational models of periodic- and aperiodic-pattern selective cells, also called grating and bar cells, respectively, are proposed. Grating cells are found in areas V1 and V2 of the visual cortex of monkeys and respond strongly to bar gratings of a given orientation and periodicity but very weakly or not at all to single bars. This non-linear behaviour, which is quite different from the spatial frequency filtering behaviour exhibited by the other types of orientation-selective neurons such as the simple cells, is incorporated in the proposed computational model by using an AND-type non-linearity to combine the responses of simple cells with symmetric receptive field profiles and opposite polarities. The functional behaviour of bar cells, which are found in the same areas of the visual cortex as grating cells, is less well explored and documented in the literature. In general, these cells respond to single bars and their responses decrease when further bars are added to form a periodic pattern. These properties of bar cells are implemented in a computational model in which the responses of bar cells are computed as thresholded differences of the responses of corresponding complex (or simple) cells and grating cells. Bar and grating cells seem to play complementary roles in resolving the ambiguity with which the responses of simple and complex cells represent oriented visual stimuli, in that bar cells are selective only for form information as present in contours and grating cells only respond to oriented texture information. The proposed model is capable of explaining the results of neurophysiological experiments as well as the psychophysical observation that the perception of texture and the perception of form are complementary processes. Received: 4 June 1996 / Accepted in revised form: 7 October 1996     

Forest above-ground volume assessments with terrestrial laser scanning: a ground-truth validation experiment in temperate,managed forests

Background and AimsQuantifying the Earth’s forest above-ground biomass (AGB) is indispensable for effective climate action and developing forest policy. Yet, current allometric scaling models (ASMs) to estimate AGB suffer several drawbacks related to model selection and uncertainties about calibration data traceability. Terrestrial laser scanning (TLS) offers a promising non-destructive alternative. Tree volume is reconstructed from TLS point clouds with quantitative structure models (QSMs) and converted to AGB with wood basic density. Earlier studies have found overall TLS-derived forest volume estimates to be accurate, but highlighted problems for reconstructing finer branches. Our objective was to evaluate TLS for estimating tree volumes by comparison with reference volumes and volumes from ASMs.MethodsWe quantified the woody volume of 65 trees in Belgium (from 77 to 2800 L; Pinus sylvestris, Fagus sylvatica, Larix decidua, and Fraxinus excelsior) with QSMs and destructive reference measurements. We tested a volume expansion factor (VEF) approach by multiplying the solid and merchantable volume from QSMs by literature VEF values.Key ResultsStem volume was reliably estimated with TLS. Total volume was overestimated by +21 % using original QSMs, by +9 % and –12 % using two sets of VEF-augmented QSMs, and by –7.3 % using best-available ASMs. The most accurate method differed per site, and the prediction errors for each method varied considerably between sites.ConclusionsVEF-augmented QSMs were only slightly better than original QSMs for estimating tree volume for common species in temperate forests. Despite satisfying estimates with ASMs, the model choice was a large source of uncertainty, and species-specific models did not always exist. Therefore, we advocate for further improving tree volume reconstructions with QSMs, especially for fine branches, instead of collecting more ground-truth data to calibrate VEF and allometric models. Promising developments such as improved co-registration and smarter filtering approaches are ongoing to further constrain volumetric errors in TLS-derived estimates.     

Cell cycle progression

In this paper we consider cell cycle models for which the transition operator for the evolution of birth mass density is a simple, linear dynamical system with a stochastic perturbation. The convolution model for a birth mass distribution is presented. Density functions of birth mass and tail probabilities in n-th generation are calculated by a saddle-point approximation method. With these probabilities, representing the probability of exceeding an acceptable mass value, we have more control over pathological growth. A computer simulation is presented for cell proliferation in the age-dependent cell cycle model. The simulation takes into account the fact that the age-dependent model with a linear growth is a simple linear dynamical system with an additive stochastic perturbation. The simulated data as well as the experimental data (generation times for mouse L) are fitted by the proposed convolution model.     

Evolution of Gene Regulatory Networks by Fluctuating Selection and Intrinsic Constraints

Various characteristics of complex gene regulatory networks (GRNs) have been discovered during the last decade, e.g., redundancy, exponential indegree distributions, scale-free outdegree distributions, mutational robustness, and evolvability. Although progress has been made in this field, it is not well understood whether these characteristics are the direct products of selection or those of other evolutionary forces such as mutational biases and biophysical constraints. To elucidate the causal factors that promoted the evolution of complex GRNs, we examined the effect of fluctuating environmental selection and some intrinsic constraining factors on GRN evolution by using an individual-based model. We found that the evolution of complex GRNs is remarkably promoted by fixation of beneficial gene duplications under unpredictably fluctuating environmental conditions and that some internal factors inherent in organisms, such as mutational bias, gene expression costs, and constraints on expression dynamics, are also important for the evolution of GRNs. The results indicate that various biological properties observed in GRNs could evolve as a result of not only adaptation to unpredictable environmental changes but also non-adaptive processes owing to the properties of the organisms themselves. Our study emphasizes that evolutionary models considering such intrinsic constraining factors should be used as null models to analyze the effect of selection on GRN evolution.     

Modeling cell interactions under flow

In this review, we summarize the current state of understanding of the processes by which leukocytes, and other cells, such as tumor cells interact with the endothelium under various blood flow conditions. It is shown that the interactions are influenced by cell-cell adhesion properties, shear stresses due to the flow field and can also be modified by the cells microrheological properties. Different adhesion proteins are known to be involved leading to particular mechanisms by which interactions take place during inflammation or metastasis. Cell rolling, spreading, migration are discussed, as well as the effect of flow conditions on these mechanisms, including microfluidic effects. Several mathematical models proposed in recent years capturing the essential features of such interaction mechanisms are reviewed. Finally, we present a recent model in which the adhesion is given by a kinetics theory based model and the cell itself is modeled as a viscoelastic drop. Qualitative agreement is found between the predictions of this model and in vitro experiments.     

Neural integration by short term potentiation

Neurophysiological studies in the oculomotor system suggest that an integrative operation is required in order to derive an eye position signal from a command signal which usually correlates with eye velocity. Several proposed models for a neural integrator are examined. All these models incorporate some form of positive feedback as a basic mechanism. Based on the performance of the models, we argue that such a scheme require extreme high precision in order to work properly. A new model based on potentiation phenomena in synaptic transmission is proposed and is shown to be free from the deficits of most previous models. The proposed model also accounts for various neural behaviors in a very natural way. A possible implementation of the model is also discussed in the context of the vestibulo-ocular reflex (VOR).     

Retrograde adaptive resonance theory based on the role of nitric oxide in long-term potentiation

Adaptive resonance theory (ART) demonstrates how the brain learns to recognize and categorize vast amounts of information by using top–down expectations and attentional focusing. ART 3, one member of the ART family, embeds the computational properties of the chemical synapse in its search process, but it converges slowly and is lack of stability when being applied in pattern recognition and analysis. To overcome these problems, Nitric Oxide (NO), which serves as a newly discovered retrograde messenger in Long-Term Potentiation (LTP), is introduced in retrograde adaptive resonance theory (ReART) model presented in this paper. In the presented model a novel search hypothesis is proposed to incorporate angle and amplitude information of an external input vector to decide whether the input matches the long-term memory (LTM) weights of an active node or not, and the embedded NO retrograde mechanism makes the search procedure a closed loop, which improves the stability and convergence speed of the transmitter releasing mechanism in a synapse. To make the model more adaptive and practical, a forgetting mechanism is built to improve the weights updating process. Experimental results indicate that the proposed ReART model achieves low error rate, fast convergence and self-organizing weights regulation. Action Editor: Christiane Linster