Search

Smart hospitals are important components of smart cities. An intelligent medical system for brain tumor segmentation is required to construct smart hospitals. To achieve intelligent brain tumor segmentation, morphological variety and serious category imbalance must be managed effectively. Conventional deep neural networks have difficulty in predicting high-accuracy segmentation images due to these issues. To solve these problems, we propose using multimodal brain tumor images combined with the UNET and LSTM models to construct a new network structure with a mixed loss function to solve sample imbalance and describe an intelligent segmentation process to identify brain tumors. To verify the practicability of this algorithm, we used the open source Brain Tumor Segmentation Challenge dataset to train and verify the proposed network. We obtained DSCs of 0.91, 0.82, and 0.80; sensitivities of 0.93, 0.85, and 0.82; and specificities of 0.99, 0.99, and 0.98 in three tumor regions, including the whole tumor ( WT ), tumor core ( TC ), and enhanced tumor ( ET ). We also compared the results of the proposed network with those of other brain tumor segmentation methods, and the results showed that the proposed algorithm could segment different tumor lesions more accurately, highlighting its potential application value in the clinical diagnosis of brain tumors.

With the advance of artificial intelligence and communication technology in the smart city, various location-based data of users can be collected via location-based social networks (LBSNs). How to make full use of these data for accurate point-of-interest (POI) recommendation is challenging because POI selection is influenced by various factors. In this article, we propose a network representation learning-enhanced multisource information (MSI) fusion model for POI recommendation in the context of LBSNs. The proposed model jointly considers various factors, including user preference, geographical influence, and social influence for a recommendation. Specifically, the social influence is modeled by performing network representation learning methods on the constructed co-visiting user networks so that the hidden complex social relationships among users can be measured automatically. Moreover, considering the significance of user preference and geographical influence, a fusion model is designed to jointly consider user preference, social influence, and geographical influence for POI recommendation. Our method is evaluated based on two publicly available data sets and extensive experimental results demonstrate that the proposed MSI fusion model outperforms several state-of-the-art algorithms for POI recommendation in terms of precision, recall, and F1.

At present, medical images have played a more and more important role in clinical treatment. Lung images provide an important reference for doctors to make a diagnosis. Especially for surgical patients, a tumor can be accurately removed based on the full cognition about its size, position, and quantity. Therefore, computer-aided diagnosis for the analysis and treatment of a lot of lung tumor images is very important. Aiming at complexity and self-adaption of image segmentation in lung tumors, this article proposed a parallel deep learning algorithm with hybrid attention mechanism for image segmentation. First, lung parenchyma was extracted via preprocessing images. Then, images were input into hybrid attention mechanism and densely connected convolutional networks (DenseNet) module, respectively, where hybrid attention mechanism consisted of a spatial attention mechanism and a channel attention mechanism. Finally, four feasible solutions were proposed for the verification through changing the convolution quantity of dense block in DenseNet. The network structure with the better performance was achieved. The experimental results prove the parallel deep learning algorithm with hybrid attention mechanism performed well in image segmentation of lung tumors, and its accuracy can reach 94.61%.

With the advancement of Internet of Things technology, we are able to collect massive people's trajectory data from various GPS services. These large amounts of trajectory records enable us to better understand human mobility patterns. Meanwhile, we are able to extract social relationships based on these digital records to provide personalized recommendation services, such as points of interests (POI) recommendation and friend recommendation. In this paper, we propose to recommend friends for taxi drivers based on vehicular trajectory records. For this purpose, we propose to construct a vehicular ad hoc network based on co-occurrence phenomenon. Furthermore, we take advantages of the network representation learning technique on the vehicular ad hoc network for learning driver vectors. Finally, potential friends are recommended based on the similarity of driver vectors. Extensive experimental results on two real-world datasets demonstrate that our proposed method has the best performance on friend recommendation compared with several state-of-the-art methods. To the best of our knowledge, this is the first attempt to recommend friends for taxi drivers based on vehicular ad hoc network representation learning.

The evaluation and optimization of reservoir operation schemes belong to a multi-objective, multi-level and multi-attribute decision-making problem. The reservoir multi-objective dispatching model generates many Pareto feasible solutions, and decision-makers often make decisions difficult. The traditional scheme selection method has the problems that the index weight is greatly affected by subjectivity, the single weight determination method is one-sided, the model calculation is complicated, and the characteristics of the evaluation index cannot be fully extracted. Therefore, this paper proposes a new method for reservoir operation plan optimization based on fuzzy optimization and convolutional neural network. First, establish the evaluation index system of the reservoir operation plan based on the fuzzy optimization theory, select the analytic hierarchy process to determine the subjective weight of the index, the entropy weight method to determine the objective weight, use the game theory to couple the subjective and objective weights, and calculate the comprehensive evaluation value of the plan through fuzzy comprehensive evaluation. Secondly, the evaluation index and comprehensive evaluation value are used as the input and output of the convolutional neural network to establish the optimal model of the reservoir operation plan. The results of the case analysis show that the research method has high accuracy and reliability, and can provide a scientific basis for reservoir operation decision-making.

We built a data-driven prediction model based on Time Series Neural Network regression model to explore the feasibility of utilizing functional connectivity (FC) to predict the continuous fatigue levels during underload task. When participants performed a 120-minute dull driving task, experimental data was recorded. The EEG and eye closure data used in this paper are from SJTU Emotion EEG Dataset, SEED. Unlike other researches, We proposed a functional brain networks method based on Phase Lag Index (PLI) to analyze EEG effectively and incorporate the temporal dependency of PERCLOS into model training. The functional connectivity corresponding to different fatigue degrees during whole experiment were analyzed. Meanwhile, taking the clustering coefficient (C) values, degree(D) and characteristic path length (L) values as input respectively while PERCLOS as output. The results suggest that brain network analysis approaches combined with Neural Network are effective to predicts individual mental fatigue during long time driving and can improve the performance of fatigue detection with a higher prediction correlation coefficient of 0.99 and a lower RMSE value of 0.01 on average.

Although promising results have been achieved in the area of traffic sign detection, little attention has been paid to text detection on traffic signs. In fact, in today’s popular driver-less automobile industry, traffic sign text which brings abundant and valuable traffic information plays an important and indispensable role. In this work, we design an effective detector for traffic sign text, whose pipeline only consists of a preprocessing module to tackle with some complex situations, a Fully Convolutional Network (FCN) in which a Scale-transfer layer is proposed to speed up the network and a simple post-processing step. Extensive experiments on the Chinese traffic sign text dataset (CTST-1600), ICDAR 2013 and MSRA-TD500 show that the proposed method has achieved the state-of-the-art results, which proves the ability of our detector on both particularity and universality applications. We collect the Chinese text-based traffic sign dataset named CTST-1600, and it can be found at https://github.com/pummi823/test/blob/master/CTST-1600 .

The approach of the complex network has well described seismic complex systems. In this paper, this is the first time three classical network construction methods for seismicity are compared. By using the same dataset from the Southern California Seismic Network, three networks are constructed. They all present the scale-free, small-world properties, a strength-degree correlation, and an assortative mixing feature. However, they show some differences in the hierarchical clustering feature. On observing the evolution results, three measures show a similar correlation with seismicity dynamics, but one measure shows a different result. These results show that different network construction methods will present some similarities and differences in network properties. This situation needs to be considered, especially when discussing a predictive indicator of seismicity.

Based on the requirement of industrial automation, this paper proposes a three-dimensional reconstruction algorithm of the dynamic environment for industrial operation environment perception, which can be applied to many different industrial fields. According to the actual situation of the inner wall of the rolling mill, a method of lidar raw data preprocessing is proposed in this paper. The original lidar data is tagged and the detailed local point cloud information of interest is isolated. Loop detection is introduced into the algorithm to establish a globally consistent point cloud graph. The overall environment modeling algorithm is constructed to realize the real-time dynamic modeling of indoor environment in the actual industrial environment operation. The method has been evaluated by real-time reconstruction of the three-dimensional model of the specified area, which provides prior information for the operation of the robotic manipulator. We compare the performance of three-dimensional with loam, using the actual industrial environment, and show that our algorithm achieves better accuracy with reduced computational expense.

This paper builds an optimal carbon-energy combined-flow (OCECF) model to optimize the carbon emission and energy losses of power grids simultaneously. A novel multi-agent cooperative reduced-dimension Q(λ) (MCR-Q(λ)) is proposed for solving the model. Firstly, on the basis of the traditional single-objective Q(λ) algorithm, the solution space is reduced effectively to shrink the size of Q-value matrices. Then, based on the concept of ant cooperative cooperation, multi-agents are used to update the Q-value matrices iteratively, which can significantly improve the updating rate. The simulation in the IEEE 118-bus system indicates that the proposed technique can decrease the convergence speed by hundreds of times as compared with conventional Q(λ), keeping high global stability, which is very suitable for dynamic OCECF in a large and complex power grid compared with other algorithms.

Regarding the continuous development of high-speed trains and the increase of running speeds, the aerodynamic design of high-speed trains has become significantly important, while reduction of drag and noise comprises a significant challenge in order to optimize aerodynamic design of high-speed trains. The design form factor of a high-speed train is highly influenced by aerodynamic aspects including aerodynamic drag, lift force, and noise. With the high-speed train as the object, the paper aims to take bionic concept as the entry point, selecting the hummingbird as the bionic prototype and extracting bionic elements to establish a bionic train model. Then, the finite volume method was used for numerical simulation and analysis of the aerodynamic performance and aerodynamic noise of the bionic high-speed train. Computational results prove that drag and noise of the bionic head type were lower than those of the original train; drag of the head train of the bionic model was reduced by 2.21% in comparison with the original model, while the whole-train drag was reduced by 3.53%, indicating that drag reduction effects are available and implying that the bionic head type could reduce drag and noise. Noise sources of the bionic train are mainly located at positions with easy airflow separation and violent turbulence motion. Large turbulence energy is in bogie areas and mainly exists at the leeward side of the bogie area. Obviously, the bogie area is the major noise source of the train. Aerodynamic noise of the bionic train in far-field comprises a wide-frequency range. Noises were concentrated within 613 Hz~3150 Hz. When the bionic high-speed train ran at 350 km/h, through comparative analysis of total noise levels at observed points of the high-speed train, it is found that this position with the maximum noise level was 25 m away from the head train nose tip, with the maximum value of 88.4 dB (A). When the bionic train ran at 600 km/h, the maximum sound pressure level at the longitudinal point was 99.7 dB (A) and the average noise level was 96.6 dB (A). When the running speed increased from 350 km/h to 600 km/h, the maximum noise level increased by 11.3 dB (A) and the average noise level increased by 11.6 dB (A). Computation results of aerodynamic noise at the point which is 7.5 m away from the rail center show that the maximum aerodynamic noise level existed at the first-end bogie of the head train, while the noise level was larger at the position closer to the ground.

Genetic Algorithm (GA) is an adaptive algorithm of global search optimization formed through the simulation of biological heredity and evolution in the natural environment. By the random selection, the algorithm requires no special needs for the search space and derivations, which is featured with simple operation, rapid convergence, and other advantages. Therefore, it is especially applicable for complex and non-linear problems that are difficult to be solved by the conventional search methods. However, this algorithm is strong in global search capability but insufficient in the local search capability. Simulated annealing (SA) is an algorithm possessed with the stronger local search ability and widely used in combinatorial optimization problems. Due to the inadequate local search capability of GA and deficient global search capability of SA, they were combined in the paper to complement their mutual advantages and take use of the global search capability of GA and local search capability of SA. The poor local search ability of GA and its premature convergence as well as the bad global search capability of SA and its low efficiency were overcome, and the SA-based mixed GA was constructed. Then, standard data sets of wine and letter-recognition in the UCI database were applied for the verification of the algorithm. It was indicated that the convergence rate was improved to some extent by the mixed algorithm proposed in this paper. Finally, the improved genetic algorithm was applied to the actual projects, which indicated the feasibility of the algorithm in engineering.

As quadruped robot is well-balanced, flexible and adaptable to environment, it has wide applications and has attracted the attention of numerous institutes. This paper combines theories of bionics and kinematics to design a quadruped robot and establish a gait control model for it. Firstly, it studies the bone structure of horse to analyze its structure advantages which are then applied to the design of the quadruped robot with the segment ratio of 1:1. Then the Kinematics formula and the traditional CPG gait control strategy are combined to establish the gait control model for quadruped robot which could output the joint control signal curves under stable gait.

In this paper, we propose an approach based on generic component model to realize the model aggregation for reconfigurable control. This is a component –oriented method. Two elementary notions, the service and the operating mode, are introduced to construct a hierarchical system and to assure the coherences between specification and realization at each level and between levels. The consideration of aggregated, complex components leads to extend this description to the possibilities of reconfiguration.

The quality of magnetic resonance imaging (MRI) features is key to the classification of Alzheimer’s disease. However, relevant research has as yet paid little attention to asymmetric MR features. In this paper, the asymmetric MR features of multiple anatomical structures are extracted. The MR feature types include volume feature and several kinds of texture features. Subsequently, the extracted features are selected based on the wrapper feature selection method with chain-like agent genetic algorithm (CAGA) and support vector machine (SVM). Finally, the selected asymmetric MR features are used for classification of Alzheimer’s disease. Experimental results show that the extracted features have apparent asymmetrical characteristics. The asymmetric volume feature of single anatomical structure can have better discrimination capability than the whole volume feature of same anatomical structure. Single selected asymmetric MR feature has displayed a superior discrimination capability in regards to three conditions of Alzheimer’s disease. The improvement is very apparent compared to before feature selection and the p-value-based feature selection method. In conclusion, this proposed method offer a new kind of feature type and can improve the classification rate for diagnosis of Alzheimer’s disease.

This paper presents a new formal modeling framework for reconfigurable control. The main idea is to use the STRIPS (STanford Research Institute Problem Solver), a formal language for automated planning which can integrate the cause-effect knowledge and the automated reasoning mechanism into one model. A qualitative model based on a graph of states is chosen. Transitions between states are determined from the possible actions (as defined as controls to the actuators). Each action is modeled in terms of preconditions and effects. The STRIPS can qualitatively define the fault models without requiring detail and precise knowledge of faulty components. Automated planning is an interesting tool for reconfigurable control. Since the process may end up in any one of a very large number of states after a break-down, it is not realistic to have recovery plans designed in advance for all such states. It is, thus, desirable to invoke a planner in the case of break-downs, which finds a plan that brings the process back into normal operation again, and that is the nature of reconfigurable control.

This paper presents a formal framework for reconfigurable control, based on model checking. This framework first generates a flexible model (i.e., an execution structure) according to the diagnosis, then defines a temporal specification language to deal with the problems due to infinite execution cycles and non-determinism, and finally provides the algorithms that will automatically verify whether the updated model satisfies the desired specification. Our entire procedure of reconfigurable control is as follows: for a given series of observations and diagnoses, if there are faults in system, then a revised system model is constructed flexibly, according to the diagnoses, observations and available system descriptions. Next, model checking is applied to verify whether or not the current system model satisfies the desired objectives. This automatic reconfigurability calculation is one of the most important steps in our framework. If the given control objectives are achievable, the system will run according to the observations and current system model. Otherwise, the reconfiguration is considered as failed. Keywords-STRIPS; Model checking; Automated Planning; Reconfigurable Control; Cause-effect

International audience; Systems are built by connecting different components (e.g., sensors, actuators, process components) that are, in turn, organized to achieve system objectives. But, when a system component fails, the system's objectives can no longer be achieved. For many years, numerous studies have proposed efficient fault detection and isolation (FDI) and fault-tolerant control (FTC) algorithms. This paper considers faults that lead to the complete failure of actuators. In this specific case, the system's physical structure changes, and the system model thus becomes incorrect. The potential that the system has to continue to achieve its objectives has to be re-evaluated from a qualitative point of view, before recalculating or modifying the control algorithms. To this end, this paper proposes a self-updating system model to reflect the current system potential, a formulation of system objectives using temporal logic, and a verification method based on model checking to verify if the objectives can still be achieved by the faulty system. The systems considered are discrete-continuous systems.

International audience; This paper continues to present the framework of multi-faults diagnosis from the preceding paper. There are two main contributions in this part. One is aimed at developing a new diagnostic process for continuous and dynamic system and its corresponding consistency-checking module. This diagnostic process and its consistency-checking module are all based on the models defined by STRIPS actions. The other is that the fault models are introduced into the consistency-checking module for preventing the impossible diagnoses. Moreover, the STRIPS can qualitatively define the fault models without requiring detail and precise knowledge about faulty components.

International audience; This paper is part I of a two part effort that is intended to present a framework of multi-faults diagnosis. Reiter [14] has proposed a consistency-based approach for multi-faults diagnosis. We extend his theory to deal with the dynamic and continuous systems and offer a necessary assumption and a formal demonstration. Multi-faults diagnosis is a partially observable problem because there is usually not enough information about faults. So the STRIPS, a classic technique of automated planning, is chosen to build the system model. It provides the reasoning ability for the multi-faults diagnosis when diagnosis is formalized as reasoning from effects to causes with causal knowledge.

This paper proposes a formal framework for reconfigurable control, based on model checking. This framework first generates a flexible model (i.e., an execution structure) according to the diagnosis, then defines a temporal specification language to deal with the problems due to infinite execution cycles and non-determinism, and finally provides the algorithms that will automatically verify whether the updated model satisfies the desired specification.

This paper proposes a generic component model (GCM) to give a functional representation of large scale systems. The GCM rests on the notion of services provided by components and their organization into operating modes. To each operating mode corresponds a behavior model uses as reference to diagnose fault. Faults are detected by local diagnosers built on subsystems using a merging procedure without any global model and global diagnoser.