Handling the massive amount of data generated by Smart Mobile Devices(SMDs)is a challenging computational problem.Edge Computing is an emerging computation paradigm that is employed to conquer this problem.It can brin...Handling the massive amount of data generated by Smart Mobile Devices(SMDs)is a challenging computational problem.Edge Computing is an emerging computation paradigm that is employed to conquer this problem.It can bring computation power closer to the end devices to reduce their computation latency and energy consumption.Therefore,this paradigm increases the computational ability of SMDs by collaboration with edge servers.This is achieved by computation offloading from the mobile devices to the edge nodes or servers.However,not all applications benefit from computation offloading,which is only suitable for certain types of tasks.Task properties,SMD capability,wireless channel state,and other factors must be counted when making computation offloading decisions.Hence,optimization methods are important tools in scheduling computation offloading tasks in Edge Computing networks.In this paper,we review six types of optimization methods-they are Lyapunov optimization,convex optimization,heuristic techniques,game theory,machine learning,and others.For each type,we focus on the objective functions,application areas,types of offloading methods,evaluation methods,as well as the time complexity of the proposed algorithms.We discuss a few research problems that are still open.Our purpose for this review is to provide a concise summary that can help new researchers get started with their computation offloading researches for Edge Computing networks.展开更多
This paper addresses the problem of predicting population density leveraging cellular station data.As wireless communication devices are commonly used,cellular station data has become integral for estimating populatio...This paper addresses the problem of predicting population density leveraging cellular station data.As wireless communication devices are commonly used,cellular station data has become integral for estimating population figures and studying their movement,thereby implying significant contributions to urban planning.However,existing research grapples with issues pertinent to preprocessing base station data and the modeling of population prediction.To address this,we propose methodologies for preprocessing cellular station data to eliminate any irregular or redundant data.The preprocessing reveals a distinct cyclical characteristic and high-frequency variation in population shift.Further,we devise a multi-view enhancement model grounded on the Transformer(MVformer),targeting the improvement of the accuracy of extended time-series population predictions.Comparative experiments,conducted on the above-mentioned population dataset using four alternate Transformer-based models,indicate that our proposedMVformer model enhances prediction accuracy by approximately 30%for both univariate and multivariate time-series prediction assignments.The performance of this model in tasks pertaining to population prediction exhibits commendable results.展开更多
Modern scintillator-based radiation detectors require silicon photomultipliers(Si PMs)with photon detection efficiency higher than 40%at 420 nm,possibly extended to the vacuum ultraviolet(VUV)region,single-photon time...Modern scintillator-based radiation detectors require silicon photomultipliers(Si PMs)with photon detection efficiency higher than 40%at 420 nm,possibly extended to the vacuum ultraviolet(VUV)region,single-photon time resolution(SPTR)<100 ps,and dark count rate(DCR)<150 kcps/mm^(2).To enable single-photon time stamping,digital electronics and sensitive microcells need to be integrated in the same CMOS substrate,with a readout frame rate higher than 5 MHz for arrays extending over a total area up to 4 mm×4 mm.This is challenging due to the increasing doping concentrations at low CMOS scales,deep-level carrier generation in shallow trench isolation fabrication,and power consumption,among others.The advances at 350 and 110 nm CMOS nodes are benchmarked against available Si PMs obtained in CMOS and commercial customized technologies.The concept of digital multithreshold Si PMs with a single microcell readout is finally reported,proposing a possible direction toward fully digital scintillator-based radiation detectors.展开更多
Feature selection plays an important role in data mining and recognition, especially in the large scale text, image and biological data. Specifically, the class label information is unavailable to guide the selection ...Feature selection plays an important role in data mining and recognition, especially in the large scale text, image and biological data. Specifically, the class label information is unavailable to guide the selection of minimal feature subset in unsupervised feature selection, which is challenging and interesting. An unsupervised feature selection based on Markov blanket and particle swarm optimization is proposed named as UFSMB-PSO. The proposed method seeks to find the high-quality feature subset through multi-particles' cooperation of particle swarm optimization without using any learning algorithms. Moreover, the features' relevance will be computed based on an information metric of relevance gain,which provides an information theoretical foundation for finding the minimization of the redundancy between features. Our results on several benchmark datasets demonstrate that UFSMB-PSO can achieve significant improvement over state of the art unsupervised methods.展开更多
The potential mechanisms of the spreading phenomena uncover the organizations and functions of various systems.However,due to the lack of valid data,most of early works are limited to the simulated process on model ne...The potential mechanisms of the spreading phenomena uncover the organizations and functions of various systems.However,due to the lack of valid data,most of early works are limited to the simulated process on model networks.In this paper,we track and analyze the propagation paths of real spreading events on two social networks:Twitter and Brightkite.The empirical analysis reveals that the spreading probability and the spreading velocity present the explosive growth within a short period,where the spreading probability measures the transferring likelihood between two neighboring nodes,and the spreading velocity is the growth rate of the information in the whole network.Besides,we observe the asynchronism between the spreading probability and the spreading velocity.To explain the interesting and abnormal issue,we introduce the time-varying spreading probability into the susceptible-infected(SI)and linear threshold(LT)models.Both the analytic and experimental results reproduce the spreading phenomenon in real networks,which deepens our understandings of spreading problems.展开更多
The industrial supply chain networks basically capture the circulation of social resource, dominating the stability and efficiency of the industrial system. In this paper, we provide an empirical study of the topology...The industrial supply chain networks basically capture the circulation of social resource, dominating the stability and efficiency of the industrial system. In this paper, we provide an empirical study of the topology of smartphone supply chain network. The supply chain network is constructed using open online data. Our experimental results show that the smartphone supply chain network has small-world feature with scale-free degree distribution, in which a few high degree nodes play a key role in the function and can effectively reduce the communication cost. We also detect the community structure to find the basic functional unit. It shows that information communication between nodes is crucial to improve the resource utilization. We should pay attention to the global resource configuration for such electronic production management.展开更多
Recent years have witnessed the proliferation of Internet of Things(IoT),in which billions of devices are connected to the Internet,generating an overwhelming amount of data.It is challenging and infeasible to transfe...Recent years have witnessed the proliferation of Internet of Things(IoT),in which billions of devices are connected to the Internet,generating an overwhelming amount of data.It is challenging and infeasible to transfer and process trillions and zillions of bytes using the current cloud-device architecture.展开更多
In silico prediction of self-interacting proteins(SIPs)has become an important part of proteomics.There is an urgent need to develop effective and reliable prediction methods to overcome the disadvantage of high cost ...In silico prediction of self-interacting proteins(SIPs)has become an important part of proteomics.There is an urgent need to develop effective and reliable prediction methods to overcome the disadvantage of high cost and labor intensive in traditional biological wet-lab experiments.The goal of our survey is to sum up a comprehensive overview of the recent literature with the computational SIPs prediction,to provide important references for actual work in the future.In this review,we first describe the data required for the task of DTIs prediction.Then,some interesting feature extraction methods and computational models are presented on this topic in a timely manner.Afterwards,an empirical comparison is performed to demonstrate the prediction performance of some classifiers under different feature extraction and encoding schemes.Overall,we conclude and highlight potential methods for further enhancement of SIPs prediction performance as well as related research directions.展开更多
In this study,an observation points‐based positive‐unlabeled learning algorithm(hence called OP‐PUL)is proposed to deal with positive‐unlabeled learning(PUL)tasks by judiciously assigning highly credible labels to...In this study,an observation points‐based positive‐unlabeled learning algorithm(hence called OP‐PUL)is proposed to deal with positive‐unlabeled learning(PUL)tasks by judiciously assigning highly credible labels to unlabeled samples.The proposed OP‐PUL algorithm has three components.First,an observation point classifier ensemble(OPCE)algorithm is constructed to divide unlabeled samples into two categories,which are temporary positive and permanent negative samples.Second,a temporary OPC(TOPC)is trained based on the combination of original positive samples and permanent negative samples and then the permanent positive samples that are correctly classified with TOPC are retained from the temporary positive samples.Third,a permanent OPC(POPC)is finally trained based on the combination of original positive samples,permanent positive samples and permanent negative samples.An exhaustive experimental evaluation is conducted to validate the feasibility,rationality and effectiveness of the OP‐PUL algorithm,using 30 benchmark PU data sets.Results show that(1)the OP‐PUL algorithm is stable and robust as unlabeled samples and positive samples are increased in unlabeled data sets and(2)the permanent positive samples have a consistent probability distribution with the original positive samples.Moreover,a statistical analysis reveals that POPC in the OP‐PUL algorithm can yield better PUL performances on the 30 data sets in comparison with four well‐known PUL algorithms.This demonstrates that OP‐PUL is a viable algorithm to deal with PUL tasks.展开更多
Indoor localization is very critical for medical care applications, e.g., the patient localization or tracking inside the building of the hospital. Traditional Radio Frequency Identification(RFID) technologies are ver...Indoor localization is very critical for medical care applications, e.g., the patient localization or tracking inside the building of the hospital. Traditional Radio Frequency Identification(RFID) technologies are very popular in this area since their cost is very low. In such technologies, each tag acts as the transmitter and the Radio Signal Strength Indicator(RSSI) information is measured from the readers. However, RSSI information suffers severely from the multi- path phenomenon. As a result, if in a very large area, the localization accuracy will be affected seriously. In order to solve this problem, we introduce Wireless Sensor Networks(WSNs) with only a few nodes, each of which acts as both transmitter and receiver. In such networks, the change of signal strength(referred as dynamic of RSSI) is leveraged to select a cluster of reference tags as candidates. Then the fi nal target location is estimated by using the RSSI relationships between the target tag and candidate reference tags. Thus, the localization accuracy and scalability are able to be improved. We proposed two algorithms, SA-LANDMARC, and COCKTAIL. Experiments show that the localization accuracy of the two algorithms can reach 0.7m and 0.45 m, respectively. Compared to most traditional Radio Frequency(RF)-based approaches, the localization accuracy is improved at least 50%.展开更多
This paper focuses on multi-modal Information Perception(IP)for Soft Robotic Hands(SRHs)using Machine Learning(ML)algorithms.A flexible Optical Fiber-based Curvature Sensor(OFCS)is fabricated,consisting of a Light-Emi...This paper focuses on multi-modal Information Perception(IP)for Soft Robotic Hands(SRHs)using Machine Learning(ML)algorithms.A flexible Optical Fiber-based Curvature Sensor(OFCS)is fabricated,consisting of a Light-Emitting Diode(LED),photosensitive detector,and optical fiber.Bending the roughened optical fiber generates lower light intensity,which reflecting the curvature of the soft finger.Together with the curvature and pressure information,multi-modal IP is performed to improve the recognition accuracy.Recognitions of gesture,object shape,size,and weight are implemented with multiple ML approaches,including the Supervised Learning Algorithms(SLAs)of K-Nearest Neighbor(KNN),Support Vector Machine(SVM),Logistic Regression(LR),and the unSupervised Learning Algorithm(un-SLA)of K-Means Clustering(KMC).Moreover,Optical Sensor Information(OSI),Pressure Sensor Information(PSI),and Double-Sensor Information(DSI)are adopted to compare the recognition accuracies.The experiment results demonstrate that the proposed sensors and recognition approaches are feasible and effective.The recognition accuracies obtained using the above ML algorithms and three modes of sensor information are higer than 85 percent for almost all combinations.Moreover,DSI is more accurate when compared to single modal sensor information and the KNN algorithm with a DSI outperforms the other combinations in recognition accuracy.展开更多
Computer clusters with the shared-nothing architecture are the major computing platforms for big data processing and analysis.In cluster computing,data partitioning and sampling are two fundamental strategies to speed...Computer clusters with the shared-nothing architecture are the major computing platforms for big data processing and analysis.In cluster computing,data partitioning and sampling are two fundamental strategies to speed up the computation of big data and increase scalability.In this paper,we present a comprehensive survey of the methods and techniques of data partitioning and sampling with respect to big data processing and analysis.We start with an overview of the mainstream big data frameworks on Hadoop clusters.The basic methods of data partitioning are then discussed including three classical horizontal partitioning schemes:range,hash,and random partitioning.Data partitioning on Hadoop clusters is also discussed with a summary of new strategies for big data partitioning,including the new Random Sample Partition(RSP)distributed model.The classical methods of data sampling are then investigated,including simple random sampling,stratified sampling,and reservoir sampling.Two common methods of big data sampling on computing clusters are also discussed:record-level sampling and blocklevel sampling.Record-level sampling is not as efficient as block-level sampling on big distributed data.On the other hand,block-level sampling on data blocks generated with the classical data partitioning methods does not necessarily produce good representative samples for approximate computing of big data.In this survey,we also summarize the prevailing strategies and related work on sampling-based approximation on Hadoop clusters.We believe that data partitioning and sampling should be considered together to build approximate cluster computing frameworks that are reliable in both the computational and statistical respects.展开更多
In this paper,an outer-rotor parallel-hybrid-excited vernier machine is proposed.In order to realize the parallel hybrid excitation,the homopolar configuration is adopted to artfully combine the permanent magnet(PM)fl...In this paper,an outer-rotor parallel-hybrid-excited vernier machine is proposed.In order to realize the parallel hybrid excitation,the homopolar configuration is adopted to artfully combine the permanent magnet(PM)flux path and the electromagnet flux path together.Firstly,the structure of the proposed machine is presented.Secondly,its operating principle is studied and discussed.Thirdly,its performance for flux regulation is analyzed based on the 3-D finite element analysis.Finally,the prototype of the proposed machine is manufactured and tested to verify the analytical results.展开更多
With the increasing pervasiveness of mobile devices such as smartphones,smart TVs,and wearables,smart sensing,transforming the physical world into digital information based on various sensing medias,has drawn research...With the increasing pervasiveness of mobile devices such as smartphones,smart TVs,and wearables,smart sensing,transforming the physical world into digital information based on various sensing medias,has drawn researchers’great attention.Among different sensing medias,WiFi and acoustic signals stand out due to their ubiquity and zero hardware cost.Based on different basic principles,researchers have proposed different technologies for sensing applications with WiFi and acoustic signals covering human activity recognition,motion tracking,indoor localization,health monitoring,and the like.To enable readers to get a comprehensive understanding of ubiquitous wireless sensing,we conduct a survey of existing work to introduce their underlying principles,proposed technologies,and practical applications.Besides we also discuss some open issues of this research area.Our survey reals that as a promising research direction,WiFi and acoustic sensing technologies can bring about fancy applications,but still have limitations in hardware restriction,robustness,and applicability.展开更多
Distributed computing frameworks are the fundamental component of distributed computing systems.They provide an essential way to support the efficient processing of big data on clusters or cloud.The size of big data i...Distributed computing frameworks are the fundamental component of distributed computing systems.They provide an essential way to support the efficient processing of big data on clusters or cloud.The size of big data increases at a pace that is faster than the increase in the big data processing capacity of clusters.Thus,distributed computing frameworks based on the MapReduce computing model are not adequate to support big data analysis tasks which often require running complex analytical algorithms on extremely big data sets in terabytes.In performing such tasks,these frameworks face three challenges:computational inefficiency due to high I/O and communication costs,non-scalability to big data due to memory limit,and limited analytical algorithms because many serial algorithms cannot be implemented in the MapReduce programming model.New distributed computing frameworks need to be developed to conquer these challenges.In this paper,we review MapReduce-type distributed computing frameworks that are currently used in handling big data and discuss their problems when conducting big data analysis.In addition,we present a non-MapReduce distributed computing framework that has the potential to overcome big data analysis challenges.展开更多
Unaided authentication services provide the flexibility to login without being dependent on any additional device.The power of recording attack resilient unaided authentication services(RARUAS)is undeniable as,in some...Unaided authentication services provide the flexibility to login without being dependent on any additional device.The power of recording attack resilient unaided authentication services(RARUAS)is undeniable as,in some aspects,they are even capable of offering better security than the biometric based authentication systems.However,high login complexity of these RARUAS makes them far from usable in practice.The adopted information leakage control strategies have often been identified as the primary cause behind such high login complexities.Though recent proposals have made some significant efforts in designing a usable RARUAS by reducing its login complexity,most of them have failed to achieve the desired usability standard.In this paper,we have introduced a new notion of controlling the information leakage rate.By maintaining a good security standard,the introduced idea helps to reduce the login complexity of our proposed mechanism—named as Textual-Graphical Password-based Mechanism or TGPM,by a significant extent.Along with resisting the recording attack,TGPM also achieves a remarkable property of threat detection.To the best of our knowledge,TGPM is the first RARUAS,which can both prevent and detect the activities of the opportunistic recording attackers who can record the complete login activity of a genuine user for a few login sessions.Our study reveals that TGPM assures much higher session resiliency compared to the existing authentication services,having the same or even higher login complexities.Moreover,TGPM stores the password information in a distributed way and thus restricts the adversaries to learn the complete secret from a single compromised server.A thorough theoretical analysis has been performed to prove the strength of our proposal from both the security and usability perspectives.We have also conducted an experimental study to support the theoretical argument made on the usability standard of TGPM.展开更多
To cater for the scenario of coordinated transportation of multiple trucks on the highway,a platoon system for autonomous driving has been extensively explored in the industry.Before such a platoon is deployed,it is n...To cater for the scenario of coordinated transportation of multiple trucks on the highway,a platoon system for autonomous driving has been extensively explored in the industry.Before such a platoon is deployed,it is necessary to ensure the safety of its driving behavior,whereby each vehicle’s behavior is commanded by the decision-making function whose decision is based on the observed driving scenario.However,there is currently a lack of verification methods to ensure the reliability of the scenario-based decision-making process in the platoon system.In this paper,we focus on the platoon driving scenario,whereby the platoon is composed of intelligent heavy trucks driving on cross-sea highways.We propose a formal modeling and verification approach to provide safety assurance for platoon vehicles’cooperative driving behaviors.The existing Multi-Lane Spatial Logic(MLSL)with a dedicated abstract model can express driving scene spatial properties and prove the safety of multi-lane traffic maneuvers under the single-vehicle perspective.To cater for the platoon system’s multi-vehicle perspective,we modify the existing abstract model and propose a Multi-Agent Spatial Logic(MASL)that extends MLSL by relative orientation and multi-agent observation.We then utilize a timed automata type supporting MASL formulas to model vehicles’decision controllers for platoon driving.Taking the behavior of a human-driven vehicle(HDV)joining the platoon as a case study,we have implemented the model and verified safety properties on the UPPAAL tool to illustrate the viability of our framework.展开更多
We propose an automatic video segmentation method based on an optimized SaliencyCut equipped with information centroid(IC)detection according to level balance principle in physical theory.Unlike the existing methods,t...We propose an automatic video segmentation method based on an optimized SaliencyCut equipped with information centroid(IC)detection according to level balance principle in physical theory.Unlike the existing methods,the image information of another dimension is provided by the IC to enhance the video segmentation accuracy.Specifically,our IC is implemented based on the information-level balance principle in the image,and denoted as the information pivot by aggregating all the image information to a point.To effectively enhance the saliency value of the target object and suppress the background area,we also combine the color and the coordinate information of the image in calculating the local IC and the global IC in the image.Then saliency maps for all frames in the video are calculated based on the detected IC.By applying IC smoothing to enhance the optimized saliency detection,we can further correct the unsatisfied saliency maps,where sharp variations of colors or motions may exist in complex videos.Finally,we obtain the segmentation results based on IC-based saliency maps and optimized SaliencyCut.Our method is evaluated on the DAVIS dataset,consisting of different kinds of challenging videos.Comparisons with the state-of-the-art methods are also conducted to evaluate our method.Convincing visual results and statistical comparisons demonstrate its advantages and robustness for automatic video segmentation.展开更多
The joint interpretation of hyperspectral images(HSIs)and light detection and ranging(LiDAR)data has developed rapidly in recent years due to continuously evolving image processing technology.Nowadays,most feature ext...The joint interpretation of hyperspectral images(HSIs)and light detection and ranging(LiDAR)data has developed rapidly in recent years due to continuously evolving image processing technology.Nowadays,most feature extraction methods are carried out by convolving the raw data with fixed-size filters,whereas the structural and texture information of objects in multiple scales cannot be sufficiently exploited.In this article,a shearlet-based structure-aware filtering approach,abbreviated as ShearSAF,is proposed for HSI and LiDAR feature extraction and classification.Specifically,superpixel-guided kernel principal component analysis(KPCA)is firstly adopted on raw HSIs to reduce the dimensions.Then,the KPCA-reduced HSI and LiDAR data are converted to the shearlet domain for texture and area feature extraction.In contrast,superpixel segmentation algorithm utilizes the raw HSI data to obtain the initial oversegmentation map.Subsequently,by utilizing a well-designed minimum merging cost that fully considers spectral(HSI and LiDAR data),texture,and area features,a region merging procedure is gradually conducted to produce a final merging map.Further,a scale map that locally indicates the filter size is achieved by calculating the edge distance.Finally,the KPCA-reduced HSI and LiDAR data are convolved with the locally adaptive filters for feature extraction,and a random forest(RF)classifier is thus adopted for classification.The effectiveness of our ShearSAF approach is verified on three real-world datasets,and the results show that the performance of ShearSAF can achieve an accuracy higher than that of comparison methods when exploiting small-size training sample problems.The codes of this work will be available at http://jiasen.tech/papers/for the sake of reproducibility.展开更多
基金supported by National Key R&D Program of China under Grant.No.2018YFB1800805National Natural Science Foundation of China under Grant No.61772345,61902257,61972261Shenzhen Science and Technology Program under Grant No.RCYX20200714114645048,No.JCYJ20190808142207420,No.GJHZ20190822095416463.
文摘Handling the massive amount of data generated by Smart Mobile Devices(SMDs)is a challenging computational problem.Edge Computing is an emerging computation paradigm that is employed to conquer this problem.It can bring computation power closer to the end devices to reduce their computation latency and energy consumption.Therefore,this paradigm increases the computational ability of SMDs by collaboration with edge servers.This is achieved by computation offloading from the mobile devices to the edge nodes or servers.However,not all applications benefit from computation offloading,which is only suitable for certain types of tasks.Task properties,SMD capability,wireless channel state,and other factors must be counted when making computation offloading decisions.Hence,optimization methods are important tools in scheduling computation offloading tasks in Edge Computing networks.In this paper,we review six types of optimization methods-they are Lyapunov optimization,convex optimization,heuristic techniques,game theory,machine learning,and others.For each type,we focus on the objective functions,application areas,types of offloading methods,evaluation methods,as well as the time complexity of the proposed algorithms.We discuss a few research problems that are still open.Our purpose for this review is to provide a concise summary that can help new researchers get started with their computation offloading researches for Edge Computing networks.
基金Guangdong Basic and Applied Basic Research Foundation under Grant No.2024A1515012485in part by the Shenzhen Fundamental Research Program under Grant JCYJ20220810112354002.
文摘This paper addresses the problem of predicting population density leveraging cellular station data.As wireless communication devices are commonly used,cellular station data has become integral for estimating population figures and studying their movement,thereby implying significant contributions to urban planning.However,existing research grapples with issues pertinent to preprocessing base station data and the modeling of population prediction.To address this,we propose methodologies for preprocessing cellular station data to eliminate any irregular or redundant data.The preprocessing reveals a distinct cyclical characteristic and high-frequency variation in population shift.Further,we devise a multi-view enhancement model grounded on the Transformer(MVformer),targeting the improvement of the accuracy of extended time-series population predictions.Comparative experiments,conducted on the above-mentioned population dataset using four alternate Transformer-based models,indicate that our proposedMVformer model enhances prediction accuracy by approximately 30%for both univariate and multivariate time-series prediction assignments.The performance of this model in tasks pertaining to population prediction exhibits commendable results.
基金supported by the National Natural Science Foundation of China(Nos.62250002,62027808,and 62027801)the Sino-German Mobility Programme(No.M-0387)。
文摘Modern scintillator-based radiation detectors require silicon photomultipliers(Si PMs)with photon detection efficiency higher than 40%at 420 nm,possibly extended to the vacuum ultraviolet(VUV)region,single-photon time resolution(SPTR)<100 ps,and dark count rate(DCR)<150 kcps/mm^(2).To enable single-photon time stamping,digital electronics and sensitive microcells need to be integrated in the same CMOS substrate,with a readout frame rate higher than 5 MHz for arrays extending over a total area up to 4 mm×4 mm.This is challenging due to the increasing doping concentrations at low CMOS scales,deep-level carrier generation in shallow trench isolation fabrication,and power consumption,among others.The advances at 350 and 110 nm CMOS nodes are benchmarked against available Si PMs obtained in CMOS and commercial customized technologies.The concept of digital multithreshold Si PMs with a single microcell readout is finally reported,proposing a possible direction toward fully digital scintillator-based radiation detectors.
基金supported by the National Natural Science Foundation of China(6113900261501229+1 种基金11547040)the Guangdong Natural Science Foundation(2016A030310051)
文摘Feature selection plays an important role in data mining and recognition, especially in the large scale text, image and biological data. Specifically, the class label information is unavailable to guide the selection of minimal feature subset in unsupervised feature selection, which is challenging and interesting. An unsupervised feature selection based on Markov blanket and particle swarm optimization is proposed named as UFSMB-PSO. The proposed method seeks to find the high-quality feature subset through multi-particles' cooperation of particle swarm optimization without using any learning algorithms. Moreover, the features' relevance will be computed based on an information metric of relevance gain,which provides an information theoretical foundation for finding the minimization of the redundancy between features. Our results on several benchmark datasets demonstrate that UFSMB-PSO can achieve significant improvement over state of the art unsupervised methods.
基金Project supported by the National Natural Science Foundation of China(Grant Nos.61703281,11547040,61803266,61503140,and 61873171)the PhD Start-Up Fund of Natural Science Foundation of Guangdong Province,China(Grant Nos.2017A030310374 and 2016A030313036)+1 种基金the Science and Technology Innovation Commission of Shenzhen,China(Grant No.JCYJ20180305124628810)the China Scholarship Council(Grant No.201806340213).
文摘The potential mechanisms of the spreading phenomena uncover the organizations and functions of various systems.However,due to the lack of valid data,most of early works are limited to the simulated process on model networks.In this paper,we track and analyze the propagation paths of real spreading events on two social networks:Twitter and Brightkite.The empirical analysis reveals that the spreading probability and the spreading velocity present the explosive growth within a short period,where the spreading probability measures the transferring likelihood between two neighboring nodes,and the spreading velocity is the growth rate of the information in the whole network.Besides,we observe the asynchronism between the spreading probability and the spreading velocity.To explain the interesting and abnormal issue,we introduce the time-varying spreading probability into the susceptible-infected(SI)and linear threshold(LT)models.Both the analytic and experimental results reproduce the spreading phenomenon in real networks,which deepens our understandings of spreading problems.
基金Project supported by the National Natural Science Foundation of China(Grant Nos.11547040 and 61703281)Guangdong Province Natural Science Foundation,China(Grant Nos.2016A030310051 and 2015KONCX143)+4 种基金Shenzhen Fundamental Research Foundation,China(Grant Nos.JCYJ20150625101524056 and JCYJ20160520162743717)SZU Student Innovation Fund,China,the PhD Start-up Fund of Natural Science Foundation of Guangdong Province,China(Grant No.2017A030310374)the Young Teachers Start-up Fund of Natural Science Foundation of Shenzhen University,Chinathe Natural Science Foundation of SZU,China(Grant No.2016-24)the Singapore Ministry of Education Academic Research Fund Tier 2(Grant No.MOE 2013-T2-2-033)
文摘The industrial supply chain networks basically capture the circulation of social resource, dominating the stability and efficiency of the industrial system. In this paper, we provide an empirical study of the topology of smartphone supply chain network. The supply chain network is constructed using open online data. Our experimental results show that the smartphone supply chain network has small-world feature with scale-free degree distribution, in which a few high degree nodes play a key role in the function and can effectively reduce the communication cost. We also detect the community structure to find the basic functional unit. It shows that information communication between nodes is crucial to improve the resource utilization. We should pay attention to the global resource configuration for such electronic production management.
文摘Recent years have witnessed the proliferation of Internet of Things(IoT),in which billions of devices are connected to the Internet,generating an overwhelming amount of data.It is challenging and infeasible to transfer and process trillions and zillions of bytes using the current cloud-device architecture.
基金This work was supported by the National Key R&D Program of China(2020YFA0908700 and 2018AAA0100100)the National Natural Science Foundation of China(Grant Nos.62002297,61902342,U1713212,61836005,and 62073225)+2 种基金the Natural Science Foundation of Guangdong Province-Outstanding Youth Program(2019B151502018)the Technology Research Project of Shenzhen City(JSGG20180507182904693)Public Technology Platform of Shenzhen City(GGFW2018021118145859).
文摘In silico prediction of self-interacting proteins(SIPs)has become an important part of proteomics.There is an urgent need to develop effective and reliable prediction methods to overcome the disadvantage of high cost and labor intensive in traditional biological wet-lab experiments.The goal of our survey is to sum up a comprehensive overview of the recent literature with the computational SIPs prediction,to provide important references for actual work in the future.In this review,we first describe the data required for the task of DTIs prediction.Then,some interesting feature extraction methods and computational models are presented on this topic in a timely manner.Afterwards,an empirical comparison is performed to demonstrate the prediction performance of some classifiers under different feature extraction and encoding schemes.Overall,we conclude and highlight potential methods for further enhancement of SIPs prediction performance as well as related research directions.
基金National Natural Science Foundation of China,Grant/Award Number:61972261Natural Science Foundation of Guangdong Province,Grant/Award Number:2314050006683+1 种基金Key Basic Research Foundation of Shenzhen,Grant/Award Number:JCYJ20220818100205012Basic Research Foundations of Shenzhen,Grant/Award Number:JCYJ20210324093609026.
文摘In this study,an observation points‐based positive‐unlabeled learning algorithm(hence called OP‐PUL)is proposed to deal with positive‐unlabeled learning(PUL)tasks by judiciously assigning highly credible labels to unlabeled samples.The proposed OP‐PUL algorithm has three components.First,an observation point classifier ensemble(OPCE)algorithm is constructed to divide unlabeled samples into two categories,which are temporary positive and permanent negative samples.Second,a temporary OPC(TOPC)is trained based on the combination of original positive samples and permanent negative samples and then the permanent positive samples that are correctly classified with TOPC are retained from the temporary positive samples.Third,a permanent OPC(POPC)is finally trained based on the combination of original positive samples,permanent positive samples and permanent negative samples.An exhaustive experimental evaluation is conducted to validate the feasibility,rationality and effectiveness of the OP‐PUL algorithm,using 30 benchmark PU data sets.Results show that(1)the OP‐PUL algorithm is stable and robust as unlabeled samples and positive samples are increased in unlabeled data sets and(2)the permanent positive samples have a consistent probability distribution with the original positive samples.Moreover,a statistical analysis reveals that POPC in the OP‐PUL algorithm can yield better PUL performances on the 30 data sets in comparison with four well‐known PUL algorithms.This demonstrates that OP‐PUL is a viable algorithm to deal with PUL tasks.
基金supported in part by China NSFC Grant 61202377 and 61170076the Guangdong Natural Science Foundation under Grant 2014A030313553+2 种基金the China National High Technology Research and Development Program 863, under Grant 2015AA015305Joint Funds of the National Natural Science Foundation of China under Grant U1301252Guangdong Province Key Laboratory Project under grant 2012A061400024
文摘Indoor localization is very critical for medical care applications, e.g., the patient localization or tracking inside the building of the hospital. Traditional Radio Frequency Identification(RFID) technologies are very popular in this area since their cost is very low. In such technologies, each tag acts as the transmitter and the Radio Signal Strength Indicator(RSSI) information is measured from the readers. However, RSSI information suffers severely from the multi- path phenomenon. As a result, if in a very large area, the localization accuracy will be affected seriously. In order to solve this problem, we introduce Wireless Sensor Networks(WSNs) with only a few nodes, each of which acts as both transmitter and receiver. In such networks, the change of signal strength(referred as dynamic of RSSI) is leveraged to select a cluster of reference tags as candidates. Then the fi nal target location is estimated by using the RSSI relationships between the target tag and candidate reference tags. Thus, the localization accuracy and scalability are able to be improved. We proposed two algorithms, SA-LANDMARC, and COCKTAIL. Experiments show that the localization accuracy of the two algorithms can reach 0.7m and 0.45 m, respectively. Compared to most traditional Radio Frequency(RF)-based approaches, the localization accuracy is improved at least 50%.
基金support provided by the National Natural Science Foundation of China (Nos. 61803267 and 61572328)the China Postdoctoral Science Foundation (No.2017M622757)+1 种基金the Beijing Science and Technology program (No.Z171100000817007)the National Science Foundation of China (NSFC) and the German Re-search Foundation (DFG) in the project Cross Modal Learning,NSFC 61621136008/DFG TRR-169
文摘This paper focuses on multi-modal Information Perception(IP)for Soft Robotic Hands(SRHs)using Machine Learning(ML)algorithms.A flexible Optical Fiber-based Curvature Sensor(OFCS)is fabricated,consisting of a Light-Emitting Diode(LED),photosensitive detector,and optical fiber.Bending the roughened optical fiber generates lower light intensity,which reflecting the curvature of the soft finger.Together with the curvature and pressure information,multi-modal IP is performed to improve the recognition accuracy.Recognitions of gesture,object shape,size,and weight are implemented with multiple ML approaches,including the Supervised Learning Algorithms(SLAs)of K-Nearest Neighbor(KNN),Support Vector Machine(SVM),Logistic Regression(LR),and the unSupervised Learning Algorithm(un-SLA)of K-Means Clustering(KMC).Moreover,Optical Sensor Information(OSI),Pressure Sensor Information(PSI),and Double-Sensor Information(DSI)are adopted to compare the recognition accuracies.The experiment results demonstrate that the proposed sensors and recognition approaches are feasible and effective.The recognition accuracies obtained using the above ML algorithms and three modes of sensor information are higer than 85 percent for almost all combinations.Moreover,DSI is more accurate when compared to single modal sensor information and the KNN algorithm with a DSI outperforms the other combinations in recognition accuracy.
基金Supported in part by the National Natural Science Foundation of China(No.61972261)the National Key R&D Program of China(No.2017YFC0822604-2)
文摘Computer clusters with the shared-nothing architecture are the major computing platforms for big data processing and analysis.In cluster computing,data partitioning and sampling are two fundamental strategies to speed up the computation of big data and increase scalability.In this paper,we present a comprehensive survey of the methods and techniques of data partitioning and sampling with respect to big data processing and analysis.We start with an overview of the mainstream big data frameworks on Hadoop clusters.The basic methods of data partitioning are then discussed including three classical horizontal partitioning schemes:range,hash,and random partitioning.Data partitioning on Hadoop clusters is also discussed with a summary of new strategies for big data partitioning,including the new Random Sample Partition(RSP)distributed model.The classical methods of data sampling are then investigated,including simple random sampling,stratified sampling,and reservoir sampling.Two common methods of big data sampling on computing clusters are also discussed:record-level sampling and blocklevel sampling.Record-level sampling is not as efficient as block-level sampling on big distributed data.On the other hand,block-level sampling on data blocks generated with the classical data partitioning methods does not necessarily produce good representative samples for approximate computing of big data.In this survey,we also summarize the prevailing strategies and related work on sampling-based approximation on Hadoop clusters.We believe that data partitioning and sampling should be considered together to build approximate cluster computing frameworks that are reliable in both the computational and statistical respects.
基金Supported in part by the National Natural Science Foundation of China under Grant 51607114in part by China Postdoctoral Science Foundation under Grant 2016M600673in part by Research Council of the University of Macao under Grant MYRG2017-00158-FST.
文摘In this paper,an outer-rotor parallel-hybrid-excited vernier machine is proposed.In order to realize the parallel hybrid excitation,the homopolar configuration is adopted to artfully combine the permanent magnet(PM)flux path and the electromagnet flux path together.Firstly,the structure of the proposed machine is presented.Secondly,its operating principle is studied and discussed.Thirdly,its performance for flux regulation is analyzed based on the 3-D finite element analysis.Finally,the prototype of the proposed machine is manufactured and tested to verify the analytical results.
基金supported by the National Natural Science Foundation of China under Grant Nos.62172286 and U2001207the Natural Science Foundation of Guangdong Province of China under Grant Nos.2022A1515011509 and 2017A030312008the Guangdong"Pearl River Talent Recruitment Program"under Grant No.2019ZT08X603.
文摘With the increasing pervasiveness of mobile devices such as smartphones,smart TVs,and wearables,smart sensing,transforming the physical world into digital information based on various sensing medias,has drawn researchers’great attention.Among different sensing medias,WiFi and acoustic signals stand out due to their ubiquity and zero hardware cost.Based on different basic principles,researchers have proposed different technologies for sensing applications with WiFi and acoustic signals covering human activity recognition,motion tracking,indoor localization,health monitoring,and the like.To enable readers to get a comprehensive understanding of ubiquitous wireless sensing,we conduct a survey of existing work to introduce their underlying principles,proposed technologies,and practical applications.Besides we also discuss some open issues of this research area.Our survey reals that as a promising research direction,WiFi and acoustic sensing technologies can bring about fancy applications,but still have limitations in hardware restriction,robustness,and applicability.
基金supported by the National Natural Science Foundation of China(No.61972261)Basic Research Foundations of Shenzhen(Nos.JCYJ 20210324093609026 and JCYJ20200813091134001).
文摘Distributed computing frameworks are the fundamental component of distributed computing systems.They provide an essential way to support the efficient processing of big data on clusters or cloud.The size of big data increases at a pace that is faster than the increase in the big data processing capacity of clusters.Thus,distributed computing frameworks based on the MapReduce computing model are not adequate to support big data analysis tasks which often require running complex analytical algorithms on extremely big data sets in terabytes.In performing such tasks,these frameworks face three challenges:computational inefficiency due to high I/O and communication costs,non-scalability to big data due to memory limit,and limited analytical algorithms because many serial algorithms cannot be implemented in the MapReduce programming model.New distributed computing frameworks need to be developed to conquer these challenges.In this paper,we review MapReduce-type distributed computing frameworks that are currently used in handling big data and discuss their problems when conducting big data analysis.In addition,we present a non-MapReduce distributed computing framework that has the potential to overcome big data analysis challenges.
基金This work is supported by the National Natural Science Foundation of China under Grant No. 61402294, the Natural Science Foun- dation of Guangdong Province of China under Grant No. S2013040012895, the Foundation for Distinguished Young Talents in Higher Education of Guangdong Province of China under Grant No. 2013LYM_0076, the Major Fundamental Research Project in the Science and Technology Plan of Shenzhen under Grant Nos. JCYJ20140828163633977 and JCYJ20160310095523765, and the Research and Development Program of Shenzhen under Grant Nos. ZDSYS20140509172959989, JSGG20150512162853495, and Shenfagai（2015）986.
文摘Unaided authentication services provide the flexibility to login without being dependent on any additional device.The power of recording attack resilient unaided authentication services(RARUAS)is undeniable as,in some aspects,they are even capable of offering better security than the biometric based authentication systems.However,high login complexity of these RARUAS makes them far from usable in practice.The adopted information leakage control strategies have often been identified as the primary cause behind such high login complexities.Though recent proposals have made some significant efforts in designing a usable RARUAS by reducing its login complexity,most of them have failed to achieve the desired usability standard.In this paper,we have introduced a new notion of controlling the information leakage rate.By maintaining a good security standard,the introduced idea helps to reduce the login complexity of our proposed mechanism—named as Textual-Graphical Password-based Mechanism or TGPM,by a significant extent.Along with resisting the recording attack,TGPM also achieves a remarkable property of threat detection.To the best of our knowledge,TGPM is the first RARUAS,which can both prevent and detect the activities of the opportunistic recording attackers who can record the complete login activity of a genuine user for a few login sessions.Our study reveals that TGPM assures much higher session resiliency compared to the existing authentication services,having the same or even higher login complexities.Moreover,TGPM stores the password information in a distributed way and thus restricts the adversaries to learn the complete secret from a single compromised server.A thorough theoretical analysis has been performed to prove the strength of our proposal from both the security and usability perspectives.We have also conducted an experimental study to support the theoretical argument made on the usability standard of TGPM.
基金supported by the National Key Research and Development Program of China under Grant No.2019YFB2102602。
文摘To cater for the scenario of coordinated transportation of multiple trucks on the highway,a platoon system for autonomous driving has been extensively explored in the industry.Before such a platoon is deployed,it is necessary to ensure the safety of its driving behavior,whereby each vehicle’s behavior is commanded by the decision-making function whose decision is based on the observed driving scenario.However,there is currently a lack of verification methods to ensure the reliability of the scenario-based decision-making process in the platoon system.In this paper,we focus on the platoon driving scenario,whereby the platoon is composed of intelligent heavy trucks driving on cross-sea highways.We propose a formal modeling and verification approach to provide safety assurance for platoon vehicles’cooperative driving behaviors.The existing Multi-Lane Spatial Logic(MLSL)with a dedicated abstract model can express driving scene spatial properties and prove the safety of multi-lane traffic maneuvers under the single-vehicle perspective.To cater for the platoon system’s multi-vehicle perspective,we modify the existing abstract model and propose a Multi-Agent Spatial Logic(MASL)that extends MLSL by relative orientation and multi-agent observation.We then utilize a timed automata type supporting MASL formulas to model vehicles’decision controllers for platoon driving.Taking the behavior of a human-driven vehicle(HDV)joining the platoon as a case study,we have implemented the model and verified safety properties on the UPPAAL tool to illustrate the viability of our framework.
基金This work was supported in part by the Major Project of the New Generation of Artificial Intelligence of National Key Research and Development Project,Ministry of Science and Technology of China under Grant No.2018AAA0102900the National Natural Science Foundation of China under Grant Nos.61572328 and 61973221+1 种基金the Natural Science Foundation of Guangdong Province of China under Grant Nos.2018A030313381 and 2019A1515011165The Hong Kong Polytechnic University under Grant Nos.P0030419 and P0030929.
文摘We propose an automatic video segmentation method based on an optimized SaliencyCut equipped with information centroid(IC)detection according to level balance principle in physical theory.Unlike the existing methods,the image information of another dimension is provided by the IC to enhance the video segmentation accuracy.Specifically,our IC is implemented based on the information-level balance principle in the image,and denoted as the information pivot by aggregating all the image information to a point.To effectively enhance the saliency value of the target object and suppress the background area,we also combine the color and the coordinate information of the image in calculating the local IC and the global IC in the image.Then saliency maps for all frames in the video are calculated based on the detected IC.By applying IC smoothing to enhance the optimized saliency detection,we can further correct the unsatisfied saliency maps,where sharp variations of colors or motions may exist in complex videos.Finally,we obtain the segmentation results based on IC-based saliency maps and optimized SaliencyCut.Our method is evaluated on the DAVIS dataset,consisting of different kinds of challenging videos.Comparisons with the state-of-the-art methods are also conducted to evaluate our method.Convincing visual results and statistical comparisons demonstrate its advantages and robustness for automatic video segmentation.
基金supported in part by the National Natural Science Foundation of China under Grant 41971300 and Grant 61901278in part by the Key Project of Department of Education of Guangdong Province under Grant 2020ZDZX3045in part by the Shenzhen Scientific Research and Development Funding Program under Grant JCYJ20180305124802421 and Grant JCYJ20180305125902403.
文摘The joint interpretation of hyperspectral images(HSIs)and light detection and ranging(LiDAR)data has developed rapidly in recent years due to continuously evolving image processing technology.Nowadays,most feature extraction methods are carried out by convolving the raw data with fixed-size filters,whereas the structural and texture information of objects in multiple scales cannot be sufficiently exploited.In this article,a shearlet-based structure-aware filtering approach,abbreviated as ShearSAF,is proposed for HSI and LiDAR feature extraction and classification.Specifically,superpixel-guided kernel principal component analysis(KPCA)is firstly adopted on raw HSIs to reduce the dimensions.Then,the KPCA-reduced HSI and LiDAR data are converted to the shearlet domain for texture and area feature extraction.In contrast,superpixel segmentation algorithm utilizes the raw HSI data to obtain the initial oversegmentation map.Subsequently,by utilizing a well-designed minimum merging cost that fully considers spectral(HSI and LiDAR data),texture,and area features,a region merging procedure is gradually conducted to produce a final merging map.Further,a scale map that locally indicates the filter size is achieved by calculating the edge distance.Finally,the KPCA-reduced HSI and LiDAR data are convolved with the locally adaptive filters for feature extraction,and a random forest(RF)classifier is thus adopted for classification.The effectiveness of our ShearSAF approach is verified on three real-world datasets,and the results show that the performance of ShearSAF can achieve an accuracy higher than that of comparison methods when exploiting small-size training sample problems.The codes of this work will be available at http://jiasen.tech/papers/for the sake of reproducibility.
