Search

Complex networks in real world are always in the state of evolution and composed of numerous overlapping communities. The discovery of overlapping communities in dynamic networks plays an important role in community detection research. In recent years, methods based on incremental clustering have become increasingly popular owing to their high efficiency. However, few of them can deal with communities that are both overlapping and dynamic. In this article, we propose an incremental clustering algorithm for discovering overlapping communities in dynamic networks. In the initial snapshot of a dynamic network, a degree-based seed selection strategy with concise and effective rules is employed to obtain stable and high-quality overlapping communities, in which the degree of nodes is the number of their neighboring nodes in the subgraph composed of free nodes. In the subsequent snapshots, a four-staged framework based on cascade information diffusion is proposed to update the communities incrementally. In this framework, a cascade information diffusion model is used to simulate the evolution of communities and then the fitness of nodes to the communities they belong to is updated based on node similarity. Experiments conducted on both real-world and artificial datasets show that the proposed algorithm can discover overlapping communities in dynamic networks effectively and outperform to the state-of-art baseline algorithms.

Compared to global community detection, local community detection aims to find communities that contain a given node. Therefore, it can be regarded as a specific and personalized community detection task. Local community detection algorithms based on modularity are widely studied and applied because of their concise strategies and prominent effects. However, they also face challenges, such as sensitivity to seed node selection and unstable communities. In this paper, a local community detection algorithm based on local modularity density is proposed. The algorithm divides the formation process of local communities into a core area detection stage and a local community extension stage according to community tightness based on the Jaccard coefficient. In the core area detection stage, the modularity density is used to ensure the quality of the communities. In the local community extension stage, the influence of nodes and the similarity between the nodes and the local community are utilized to determine boundary nodes to reduce the sensitivity to seed node selection. Experimental results on real and artificial networks demonstrated that the proposed algorithm can detect local communities with high accuracy and stability.

Multi-speed transmissions can enhance the comprehensive performance of electric vehicles (EVs), while exhibiting potential to further improve their shift quality. In this study, a gearshift control...

Aspect-based sentiment analysis is a challenging subtask of sentiment analysis, which aims to identify the sentiment polarities of the given aspect terms in sentences. Previous studies have demonstrated the remarkable progress achieved by memory networks. However, current memory-network-based models cannot fully exploit long-term semantic relationships to the given aspect terms in sentences, which may lead to the loss of aspect information. In this paper, we propose a novel memory network with hierarchical multi-head attention (MNHMA) for aspect-based sentiment analysis. First, we introduce a semantic information extraction strategy based on the rotational unit of memory to acquire long-term semantic information in context and build memory for the memory network. Second, we propose a hierarchical multi-head attention mechanism to preserve aspect information and enable MNHMA to focus on the critical context words to the given aspect terms in sentences. Third, we employ a fully connected layer in each attention layer of the hierarchical multi-head attention layer to simulate the nonlinear transformation of sentiments, thereby acquiring a comprehensive context representation for aspect-level sentiment classification. Experimental results on three commonly used benchmark datasets demonstrate that our MNHMA model outperforms other state-of-the-art models for aspect-based sentiment analysis.

Electric vehicles are being increasingly adopted worldwide to support the sustainable development of society. However, various technologies that are related to electric vehicles still require further investigation. In this study, we consider a gearshift control architecture that combines offline trajectory planning and online control for improved shifting in multispeed transmission electric vehicles. For gearshift trajectory planning, we establish a dynamic model and apply the multi-objective optimization Radau pseudospectral method. Simulations of this method provide a Pareto solution set under three optimization objectives, namely duration, friction work, and jerk, thereby establishing a new approach for satisfying the requirements for shift quality. Moreover, the relations among these objectives are detailed on the Pareto solution set. We conducted a hardware-in-the-loop simulation to verify the performance of the proposed control architecture and trajectory planning method. The simulation results indicate that the gearshift control delivers a suitable response for different driving intentions based on the obtained Pareto solution set.

Stereo matching is a fundamental and challenging task which has various applications in autonomous driving, dense reconstruction and other depth related tasks. Contextual information with discriminative features is crucial for accurate stereo matching in the ill-posed regions (textureless, occlusion, etc.). In this paper, we propose an efficient horizontal attention module to adaptively capture the global correspondence clues. Compared with the popular non-local attention, our horizontal attention is more effective for stereo matching with better performance and lower consumption of computation and memory. We further introduce a deformable module to refine the contextual information in the disparity discontinuous areas such as the boundary of objects. Learning-based method is adopted to construct the cost volume by concatenating the features of two branches. In order to offer explicit similarity measure to guide learning-based volume for obtaining more reasonable unimodal matching cost distribution we additionally combine the learning-based volume with the improved zero-centered group-wise correlation volume. Finally, we regularize the 4D joint cost volume by a 3D CNN module and generate the final output by disparity regression. The experimental results show that our proposed HDA-Net achieves the state-of-the-art performance on the Scene Flow dataset and obtains competitive performance on the KITTI datasets compared with the relevant networks.

Community structure is an important characteristic of complex networks. Uncovering communities in complex networks is currently a hot research topic in the field of network analysis. Local community detection algorithms based on seed-extension are widely used for addressing this problem because they excel in efficiency and effectiveness. Compared with global community detection methods, local methods can uncover communities without the integral structural information of complex networks. However, they still have quality and stability deficiencies in overlapping community detection. For this reason, a local community detection algorithm based on internal force between nodes is proposed. First, local degree central nodes and Jaccard coefficient are used to detect core members of communities as seeds in the network, thus guaranteeing that the selected seeds are central nodes of communities. Second, the node with maximum degree among seeds is pre-extended by the fitness function every time. Finally, the top k nodes with the best performance in pre-extension process are extended by the fitness function with internal force between nodes to obtain high-quality communities in the network. Experimental results on both real and artificial networks show that the proposed algorithm can uncover communities more accurately than all the comparison algorithms.

Community detection is an important research direction in the field of complex network analysis. It aims to discover community structures in complex networks. Algorithms based on dynamic distance mechanism can find stable communities with various shapes. However, they still cannot discover overlapping or outlier communities. This paper proposes an overlapping community discovery algorithm based on coarsening and local overlapping modularity. First, to reduce the running time, a new equation for computing the local overlapping modularity increment is derived. This equation finds the overlapping communities, accurately and quickly. Second, a new similarity measuring strategy is designed to reduce the number of outlier communities. The experiments on artificial and real datasets show that the proposed algorithm can discover the overlapping communities, accurately and efficiently.

Autism Spectrum Disorder (ASD) is a widely prevalent neurodevelopmental disorder with symptoms of social interaction and communication problems and restrictive and repetitive behavior. In recent years, there has been increasing interest in identifying individuals with ASD patients from typical developing (TD) ones based on brain imaging data such as MRI. Although both traditional machine learning and recent deep learning methodologies have achieved promising performance, the classification accuracy is still far from satisfying due to large individual differences and/or heterogeneity among data from different sites. To help resolve the problem, we proposed a novel Attention-based Node-Edge Graph Convolutional Network (ANEGCN) to identify ASD from TD individuals. Specifically, it simultaneously models the features of nodes and edges in the graphs and combines multi-modal MRI data including structural MRI and resting state functional fMRI in order to utilize both structural and functional information for feature extraction and classification. Moreover, an adversarial learning strategy was used to enhance the model generalizability. A gradient-based model interpretability method was also applied to identify those brain regions and connections contributing to the classification. Using the worldwide Autism Brain Imaging Data Exchange I (ABIDE I) dataset with 1007 subjects from 17 sites, the proposed ANEGCN achieved superior classification accuracy (72.7%) and generalizability than other state-of-the-art models. This study provided a powerful tool for ASD identification.

One of the most prominent anatomical characteristics of the human brain lies in its highly folded cortical surface into convex gyri and concave sulci. Previous studies have demonstrated that gyri and sulci exhibit fundamental differences in terms of genetic influences, morphology and structural connectivity as well as function. Recent studies have demonstrated time-frequency differences in neural activity between gyri and sulci. However, the functional connectivity between gyri and sulci is currently unclear. Moreover, the regularity/variability of the gyro-sulcal functional connectivity across different task domains remains unknown. To address these two questions, we developed a novel anatomy-guided spatio-temporal graph convolutional network (AG-STGCN) to classify task-based fMRI (t-fMRI) and resting state fMRI (rs-fMRI) data, and to further investigate gyro-sulcal functional connectivity differences across different task domains. By performing seven independent classifications based on seven t-fMRI and one rs-fMRI datasets of 800 subjects from the Human Connectome Project, we found that the constructed gyro-sulcal functional connectivity features could satisfactorily differentiate the t-fMRI and rs-fMRI data. For those functional connectivity features contributing to the classifications, gyri played a more crucial role than sulci in both ipsilateral and contralateral neural communications across task domains. Our study provides novel insights into unveiling the functional differentiation between gyri and sulci as well as for understanding anatomo-functional relationships in the brain.

It has been widely demonstrated that complex brain function is mediated by the interaction of multiple concurrent brain functional networks, each of which is spatially distributed across specific brain regions in a temporally dynamic fashion. Therefore, modeling spatio-temporal patterns of those holistic brain functional networks provides a foundation for understanding the brain. Compared to conventional modeling approaches such as correlation, general linear model, and matrix decomposition methods, recent deep learning methodologies have shown a superior performance. However, the existing deep learning models either underutilized both spatial and temporal characteristics of fMRI during model training, or merely focused on modeling only one targeted brain functional network at a time while ignoring holistic ones, resulting in a significant gap in our current understanding of how the brain functions. To bridge this gap, we propose a novel Multi-Head Guided Attention Graph Neural Network (Multi-Head GAGNN) to simultaneously model spatio-temporal patterns of multiple brain functional networks. In Multi-Head GAGNN, the spatial patterns of multiple brain networks are firstly modeled in a multi-head attention graph U-net, and then adopted as guidance for modeling the corresponding temporal patterns of multiple brain networks in a temporal multi-head guided attention network model. Results based on two task fMRI datasets from the public Human Connectome Project demonstrate superior ability and generalizability of Multi-Head GAGNN in simultaneously modeling spatio-temporal patterns of holistic brain functional networks compared to other state-of-the-art models. This study offers a new and powerful tool for helping understand complex brain function.

Clustering algorithms aim at gathering similar data points from a dataset in an unsupervised manner. Although the batch clustering algorithms have relatively high accuracy, they cannot make use of the dynamic clustering results efficiently. The requirement of using the whole dataset in calculation results in the problems of resource waste and high time cost. On the contrary, incremental clustering only needs to update the varied part of a model upon the arrival of new data, which makes it unnecessary to recluster the whole data all the time. The feature is very suitable for the streaming data process, but it decreases the accuracy of the algorithms and cannot satisfy the low latency requirement of real-time data processing. In response to this problem, the paper proposes a novel unified batch and streaming clustering model (UBSCM) based on streaming computation, which includes a streaming cluster feature updating mechanism (SCFUM). The Flink framework is used to implement a new streaming KMeans algorithm based on UBSCM (KMeansUBSP). The experiments on the real-world datasets validate that the new streaming KMeans algorithm is effective in clustering the batch and streaming data in a unified manner.

Wavelength-dependent light absorption and scattering will reduce the quality of underwater images. Therefore, the characteristics of underwater images are different from those taken in natural. Low-quality underwater images affect the accuracy of pattern recognition, visual understanding, and key feature extraction in underwater scenes. In this paper, we enhance the underwater image using a multi-scale generative adversarial network with adjacent scale feature addition. Adjacent scale feature addition allows the network to more effectively capture the relevant characteristics between two image domains. The multi-scale discriminator can let the enhanced image more closer to the natural image. Our method does not rely on transmission maps and atmospheric light estimation. Experiments on a large amount of synthetic data and real data show that our method is superior to the state-of-the-art methods.

Since the complex brain functions are achieved by the interaction of functional brain networks with the specific spatial distributions and temporal dynamics, modeling the spatial and temporal patterns of functional brain networks based on 4D fMRI data offers a way to understand the brain functional mechanisms. Matrix decomposition methods and deep learning methods have been developed to provide solutions. However, the underlying nature of functional brain networks remains unclear due to underutilizing the spatio-temporal characteristics of 4D fMRI input in previous methods. To address this problem, we propose a novel Guided Attention 4D Convolutional Neural Network (GA-4DCNN) to model spatial and temporal patterns of functional brain networks simultaneously. GA-4DCNN consists of two subnetworks: the spatial 4DCNN and the temporal Guided Attention (GA) network. The 4DCNN firstly extracts spatio-temporal characteristics of fMRI input to model the spatial pattern, while the GA network further models the corresponding temporal pattern guided by the modeled spatial pattern. Based on two task fMRI datasets from the Human Connectome Project, experimental results show that the proposed GA-4DCNN has superior ability and generalizability in modeling spatial and temporal patterns compared to other state-of-the-art methods. This study provides a new useful tool for modeling and understanding brain function.

Drop-and-pull transportation can repeatedly utilize tractors with different trailers and reduce costs, carbon emissions, and the number of tractors to purchase and use. Fuel-cell electric vehicles (FCEV) possess high power and long drive endurance. These performance characteristics complement the performance requirements of drop-and-pull transportation of heavy loads and long mileage. This paper proposes a novel fuel-cell electric articulated vehicle featuring three power sources: fuel cell, power battery, and ultracapacitor. Then, based on the proposed vehicle, we expound on a highly efficient and flexible transport system. To compare economics and durability of fuel-cell electric trailers with two energy sources (i.e., fuel-cell-battery) and three energy sources, we developed and simulated a rule-based energy management strategy under driving conditions. The results indicate that, although similar levels of fuel economy and capacity degradation of the fuel cell occur for the proposed vehicle and its two-energy-source counterpart, the ampere–hour throughput of three-energy-source vehicles was 64% lower than that of two-energy-source vehicles, which indicates the introduction of the ultracapacitor in fuel-cell-battery electric articulated vehicles can offer significant protection to the power battery. This result shows that the three energy sources increase the service life of the energy system.

Over the last decade various compressive snapshot hyperspectral imaging methods have been proposed. The limited reconstruction quality from severely compressed measurements, however, has been a practical barrier to real applications. This paper proposes a compressive chromo-tomography framework that incorporates the convolutional sparse coding (CSC) prior into the classical total variation and L 1 regularization functionals. Such a combination allows excellent high-frequency recovery capabilities of CSC, while effectively suppressing ghost artifacts in tomographic reconstructions. Since nondifferentiable regularizers are employed, we propose a preconditioned alternating direction method of multipliers (ADMM) for flexible and efficient solutions, both for the reconstruction task and for hyperspectral convolutional dictionary learning. We demonstrate in our numerical experiments that just 25 learned 3D CSC filters can fulfill a rather effective hyperspectral imagery representation and that the proposed method is capable of high accuracy reconstructions.

The existence of communities in various complex networks is ubiquitous in all aspects of people’s living. Hence, it is crucial to uncover communities accurately, which is one of the hottest research areas in the field of network analysis. Particularly, complex networks are usually in continuous change so that it is more realistic to uncover dynamic communities. In this study, an algorithm based on node contribution for uncovering dynamic communities is proposed. Firstly, the seed nodes are selected via node local fitness in the network, thus guaranteeing that the selected seeds are central nodes of communities. Secondly, a static algorithm is used to obtain communities in initial snapshot of the network. Finally, node contribution is proposed to incrementally uncover communities in non-initial snapshots of the network. The experimental results reveal that our method outperforms all other comparison algorithms in both artificial and real datasets.

With the occurrence of large-scale human trajectories, which imply spatial and temporal patterns, the subject of mobility prediction has been widely studied. A number of approaches are proposed to predict the next location of a user. In this paper, we expect to lengthen the temporal dimension of prediction results beyond one hop. To predict the future locations of a user at every time unit within a specified time, we propose a Markov-based multi-hop mobility prediction (Markov---MHMP) algorithm. It is a hybrid approach that considers multiple factors including personal habit, weekday similarity, and collective behavior. On a GPS dataset, our approach performs prediction better than baseline and state-of-the-art approaches under several evaluation criteria.

Discovering communities in large-scale social networks efficiently and accurately is one of the challenges in social network data mining. We propose a clustering algorithm to discover social network communities based on the propagation of local and global information. Three strategies, namely, localizing propagation of affinity messages, relaxing self-exemplar constraints, and hierarchical processing, are employed in the algorithm to achieve reasonable time and space complexities in social networks. The local and global information is represented by the k-path edge centrality incorporated in the similarity calculation. The standalone algorithm is extended to provide parallel implementations based on the MapReduce model to accelerate processing in large-scale networks. Two well-known parallel computation frameworks, Hadoop and Spark, are adopted to implement the parallel algorithm. Experiments performed on artificial and real social network datasets show that the proposed algorithms can achieve near-linear time and space complexities with comparative clustering accuracy.

With the rapid development of Virtual Reality (VR) and Augmented Reality (AR), the rendering technology has made great progress. In large games and many animated movies, the rendering technology has a wide range of applications. This paper presents a real-time rendering method includes the depth of field (DoF) and soft shadow effect, using Mipmap to simulate DoF effect, using the parallel-split shadow map (PSSM) method combined with variance shadow map (VSM) to produce soft shadows, and the Gauss function forming the DoF effect of the shadow. For Mipmap, its level can be determined according to the size of the circle of confusion (CoC). The Gauss function is used to generate the DoF effect of the shadow and its standard deviation is also determined by the CoC.

Application of cloud computing technologies in power system has made a great contribution to the establishment of smart grid. Among applications of smart grid, electrical load prediction plays an important role in efficient use of power resource. However, the exponential growth of data has posed a great challenge to the existing algorithms. In this paper, we firstly propose a novel parallel hybrid algorithm, combining the Improved Particle Swarm Optimization (PSO) with ELM, named PIPSO-ELM. Here a modified particle swarm optimization is presented to find the optimal number of hidden neurons as well as the corresponding input weights and hidden biases. Furthermore, in the iterative search process of PSO, an update strategy employs the mutation operator of evolutionary algorithms is introduced for further improving the global search capability and convergence speed of PSO. After that, to handle the large-scale dataset efficiently, the parallel implementation of PIPSO-ELM is achieved using Spark. Finally, extensive experiments on real-life electrical load data and comprehensive evaluation are conducted to verify the performance of PIPSO-ELM in electrical load prediction. Extensive experimental results demonstrate that PIPSO-ELM outperforms the compared algorithms in terms of stability, efficiency and scalability simultaneously.

Saliency maps computed by each salient object detection algorithm show similar effect against ground truth. Therefore this paper proposes to optimize the saliency value of each pixel based on the saliency values of its neighbouring pixels. It proposes a machine learning-based method to learn the relationship between ground truth and saliency maps and among saliency values of neighbouring pixels. Then it applies the learned models to a new saliency map and then confines the fitted values to the legitimate range to obtain the optimized saliency map. The optimized saliency map is closer to the ground truth and better than the original saliency map. The experimental results show that the values of many saliency evaluation metrics improved significantly and some of them remained unchanged. They illustrate that the proposed algorithm is suitable for optimizing many salient object detection algorithms.

One of the traditional ways for detecting dynamic communities is to find the communities at each interval through the static community detection algorithms. However, it usually leads to high computation complexity. In this paper, a novel algorithm based on the MapReduce model and the label propagation progress with the strategy of incremental related vertices is proposed, which is called PLPIRV (Parallel Label Propagation and Incremental Related Vertices). Based on the communities found at the previous interval, the new algorithm adjusts the communities the incremental related vertices belong to. The clustering of the whole network can be avoided by incrementally analyzing the variation of the networks, so that the time cost can be greatly reduced. Experiments on artificial and real datasets show that the proposed algorithm performs well on dynamic community detection. (Abstract).

This paper presents a novel method for scanning duplex-printed documents without incurring the unwanted show-through artifact. The proposed method achieves the goal of eliminating the leaked-out reverse-side content by fusing a white backed scan image with a black backed scan image of the document. The fusion is accomplished using a multilayer perceptron having learned a fusion mapping from manually corrected document images. The main novel contributions of this work include (1) being the first to propose to accomplish the goal of show through free scanning by fusing a white backed scan image with a black backed scan image of the document; (2) proposing a learning approach using a multilayer perceptron to learn the fusion mapping from manually corrected scan images; and (3) proposing to use the pixel value histogram of reverse-side-printed area as well as the pixel value histogram of duplex-printed area to quantitatively indicate show through severity to facilitate objective comparison of the methods in consideration. The experiment results show that the proposed method is remarkably more powerful in eliminating show through than the two state-of-the-art methods in comparison.

Associative data mining is the research hotspot in the field of big data, and frequent item sets mining is an important step in the analysis of associative data. This paper focuses on analyzing the frequent item sets mining algorithm based on Apriori parallel algorithm. The paper has found two shortages of Apriori parallel algorithm: one is that the key value pair are too many, another is that in the combiner stage, it occupies two much memory. Therefore, we propose an optimized algorithm. In the optimization algorithm, candidate item sets and local count information are saved in memory, greatly reducing the number of generated keys. Meanwhile, in the short length frequent item sets mining, the method of reducing the number of scanning transaction data without generating candidate item sets can improve the algorithm efficiency. We do the experiments in the Hadoop platform to testify the performance of the proposed optimized algorithm. The experiments demonstrate that the time and I/O of the optimized algorithm have been improved greatly, compared with the non-optimized algorithm.

Urban impedance is an important consideration in assessments of transportation and land-use systems. This work leverages check-in records obtained from mobile social networks to build a fine-grained but inexpensive urban impedance model. Check-in records and road networks are collected and used to calculate and adjust the various parameters of the model, including path length, number and angle of turns, number and direction of junctions, and population density. Check-in records can filter functional locations and supply the time factor, thereby providing excellent advantages over traditional models that do not employ this data type. The proposed model is more accurate than traditional impedance models, as verified by experiments using Sina Weibo data in Tianjin City.

Transmission Control Protocol (TCP) optimization in Mobile Ad hoc NETworks (MANETs) is a challenging issue because of some unique characteristics of MANETs. In this paper, a new end-to-end mechanism based on multiple metrics measurement is proposed to improve TCP performance in MANETs. Multi-metric Measurement based Enhancement of TCP (MME-TCP) designs the metrics and the identification algorithm according to the characteristics of MANETs and the experiment results. Furthermore, these metrics are measured at the sender node to reduce the overhead of control information over networks. Simulation results show that MME-TCP mechanism achieves a significant performance improvement over standard TCP in MANETs.

Micro-blog has become a symbol of the novel social media, and because of its rapid development in such a short time, many research researchers are full of enthusiasm about it. We take use of Latent Dirichlet Allocation (LDA) Model which has excellent dimension reduction capability and can excavate latent semantic from texts to discover popular topics. We improve the original LDA model to FSC-LDA model by combining the text clustering methods and feature selection methods, which can identify the number of topics adaptively. FSC-LDA model can keep short micro-blog texts features better, and make the result more stable. The result of the experiments on real Chinese microblog text dataset shows that FSC-LDA model can perform well on the custom evaluation and find more accurate popular topics.

Vehicular ad hoc networks (VANETs) have become important components of metropolitan area networks, and clustering for VANETS provides many advantages. However, the stability of current clustering algorithms exhibits poor robustness because a VANET is a highly dynamic scenario. In this study, a novel multi-hop clustering scheme for VANETs, which generates cluster heads (CHs) via neighborhood follow relationship between vehicles, is proposed. The scheme is based on a reasonable assumption that a vehicle cannot certainly identify which vehicle in its multi-hop neighbors is the most suitable to be its CH, but it can easily grasp which vehicle in one-hop distance is the most stable and similar with it, and thus, they most likely belong to the same cluster. Consequently, a vehicle can choose its CH by following the most stable vehicle. The relative mobility between two vehicles combining the gains based on the followed number and the historical following information enables a vehicle to select which target to follow. Extensive simulation experiments are conducted to validate the performance of the proposed clustering scheme.

Finding communities in networks is one of the challenging issues in complex network research. We have to deal with very large networks that contain billions of vertices, which makes community discovery a computationally intensive work. Moreover, communities usually overlap each other, which greatly increases the difficulty of identifying the boundaries of communities. In this paper, we propose a parallel multi-label propagation algorithm (PMLPA) that enhances traditional multi-label propagation algorithm (MLPA) in two ways. First, the critical steps of MLPA are parallelized based on the MapReduce model to get higher scalability. Second, new label updating strategy is used to automatically determine the most valuable labels of each vertex. Furthermore, we study the improvement of PMLPA through considering the influence of vertices and labels on label updating. In this way, the importance of each label can be described with higher precision. Experiments on artificial and real networks prove that the proposed algorithms can achieve both high discovering accuracy and high scalability.

Multi-label propagation algorithms (MLPAs) have nearly linear time complexity, but the accuracy and stability still need to be improved when applied to overlapping community discovery. Inspired from the idea that boundary nodes are more probable to appear in the overlapping regions of different communities, a Hierarchical Multi-label Propagation Algorithm (HMPA) based on node hierarchy and label propagation gain for overlapping community discovery in social networks is proposed in this paper. HMPA consists of three stages. Firstly, HMPA utilizes LPAm to unfold initial non-overlapping communities. Secondly, a PageRank-like method is proposed to mark the hierarchy of each node according to the initial partition of the first stage. Finally, multi-label propagation algorithm considering label propagation gain between nodes, which is calculated based on node hierarchy, is introduced to refine overlapping region. Experimental results on both the synthetic and real world networks show that the proposed algorithm can effectively solve the problems of traditional multi-label propagation algorithms in terms of accuracy and stability.

The research of topic trend prediction can be a good reference for maximizing the propagation effects of network advertisements as well as guiding and controlling the network consensus. This paper proposes PTEP(the Prediction of Topic Energy Peak) method to model the life cycle of a topic and predicts the time when a hot topic will outbreak. Firstly, taking the number and the authority of followers and the interest of users to a topic into consideration, we design a topic-related user authority (TRUA) algorithm to measure the authority of users. Secondly, we calculate the energy value considering both the tweets and users authority related to the topic. Thirdly, we measure the fluctuation of the energy value based on wavelet transformation. Finally, we present rules to predict topic trend. Experimental results show that our method can effectively predict the peak of a topic in advance with a low omission rate.

Social network analysis has been one of the research focuses with the developing of social media services. Social influence also attracts ever-increasing attention and interests from both the sociology and the data mining scholars, but the research of influence is mainly limited in the user level, the community level influence is rarely involved. In this paper, we develop a novel model to analyze and evaluate the community level influence and help the decision maker to comprehend the disparity in influence of different communities. This paper also proposes a simple heuristic node-selection strategy considering the community influence to spread influence, and the experiments of the spread of influence through the social network datasets prove the rationality of the proposed community influence analysis model.

In recent years, online social networks such as Facebook, Twitter and Sina Weibo are more and more popular and there is a highly increasing interest of studying the relationships among the large amount of microblogging users. In this paper, the link prediction method is utilized to analyze the relationships between Sina Weibo users. Firstly, the topological features of Weibo network are studied and the influence of topological structure features to the formation of Sina Weibo network is verified. Then the attribute features of Sina Weibo are also considered and analyzed. In this paper, a link prediction model is introduced based on Bayesian networks classifier combining the two types of features together. The experiments are conducted with the datasets crawled from Sina Weibo site. We compare the experiment results with and without the attribute features and rank the importance of features, finding out that the attribute features have a significant effect on the formation of Weibo users' relationships besides the topological structure features, and contribute significantly to the improvement of the predictive performance.

The service-oriented architecture is considered as a new emerging trend for the future of wireless sensor networks in which different types of sensors can be deployed in the same area to support various service requirements. The accuracy of the sensed data is one of the key criterions because it is generally a noisy version of the physical phenomenon. In this paper, we study the node selection problem with data accuracy guarantee in service-oriented wireless sensor networks. We exploit the spatial correlation between the service data and aim at selecting minimum number of nodes to provide services with data accuracy guaranteed. Firstly, we have formulated this problem into an integer nonlinear programming problem to illustrate its NP-hard property. Secondarily, we have proposed two heuristic algorithms, namely, Separate Selection Algorithm (SSA) and Combined Selection Algorithm (CSA). The SSA is designed to select nodes for each service in a separate way, and the CSA is designed to select nodes according to their contribution increment. Finally, we compare the performance of the proposed algorithms with extended simulations. The results show that CSA has better performance compared with SSA.

Community detection in social networks is usually considered as an objective optimization problem. Due to the limitation of the objective function, the global optimum cannot describe the real partition well, and it is time consuming. In this paper, a novel PSO (particle swarm optimization) algorithm based on modularity optimization for community detection in social networks is proposed. Firstly, the algorithm takes similarity-based clustering to find core areas in the network, and then a modified particle swarm optimization is performed to optimize modularity in a new constructed weighted network which is compressed from the original one, and it is equivalent to optimize modularity in the original network with some restriction. Experiments are conducted in the synthetic and four real-world networks. The experimental results show that the proposed algorithm can effectively extract the intrinsic community structures of social networks.

The performance of wireless mesh network is influenced significantly by the link state and interference between links. In this paper we seek routes for QoS connection requests with bandwidth requirements in consideration of the link performance parameters of data link layer and QoS requirements. In order to get the solution of QoS routing problem, the problem is converted to multi-objective combinatorial optimization problem and a discrete PSO algorithm is proposed for multi-objective optimization to realize the QoS routing problem. Experiment results show that compared with enumeration method and GA algorithm, our scheme for QoS routing is feasible and efficient.

With the rapid development of mobile SMS (short message service), spam messages have grown explosively which trouble our daily lives seriously and lead to the loss of telecom operators. In this paper, an online spam filter based on the analysis of two criteria of content representations and relationship between the senders and receivers in social network is proposed. A Naive Bayesian classifier is used to build up the filter including both the content features and social network features. We use the data provided by a partner telecom operator to do the experiments. The results show that our model is effective and satisfies all the requirements of our partner and will be deployed recently.

This paper describes a vehicle-borne mapping system which combines laser scanners, cameras and navigation unit for urban spatial object extraction especially the traffic signs/signals. Considering the high cost of searching candidates of urban spatial objects in images, candidates are searched in laser points. As laser scanners and cameras are calibrated into a common reference frame, laser points can be projected onto images to perform candidate verification. The combination of laser scanners and cameras can reduce the search space in images. Moreover laser points provide precise 3d position for the candidate.

TCP optimization in mobile ad hoc networks (MANETs) is a challenging issue because of some unique characteristics of MANETs. In this paper, an end-to-end mechanism based on multiple metrics measurement is proposed to improve TCP performance in mobile ad hoc networks. The metrics and the identification algorithm are designed elaborately according to the experiment results and analysis. Furthermore, these metrics are measured at the sender side to reduce the overhead of control information over network. Simulation results show that our mechanism achieves a significant performance improvement over standard TCP in MANETs

IEEE 802.11a is a popular high-speed WLAN standard with a data rate up to 54 Mbps. The fundamental medium access control (MAC) mechanism of IEEE 802.11a is the contention-based distributed coordination function (DCF) which includes two packet transmission schemes, namely, the basic access and RTS/CTS mechanism. In this paper, by introducing a two-dimensional Markov model, we conduct an extensive analysis of saturation throughput and access delay of these two access mechanisms in IEEE 802.11a WLAN. The study reveals that, in contradiction to the situation in IEEE 802.11 WLAN, basic access mechanism is always more efficient than RTS/CTS mechanism in high-speed IEEE 802.11a WLAN

This paper develops a novel scheme for Traffic Engineering (TE). In the main scheme, Differentiated Services (DiffServ) is employed to meet different QoS requirements, while load balancing is preserved. Simulation results illustrate that the proposed algorithm outperforms the traditional methods in terms of load balancing and traffic performance, so that the capability of the network is improved in evidence.