摘要:An approximately optimal subspace pursuit algorithm under cosparse analysis model was studied to reconstruct the original signal from the noisy measurement vectors.To overcome the drawbacks of the non steepest gradient during the pursuit process and the low successful reconstruction probability for sparse synthesis model,an approximately optimal subspace pursuit algorithm based on cosparse analysis model was presented and realized.The approximately optimal optimization object function for the algorithm was designed according to the structure of the different analysis dictionaries,the iterative pursuit process of the algorithm was revised,and the methods of selecting cosparsity was optimized.The simulation experiments show that the complete reconstruction probability of the new algorithm is evidently larger than that of the algorithm for ASP,AHTP,AIHT,AL1 and GAP when the cosparse operator is a random compact frame or a two dimension total variant matrix.The comprehensive average PSNR of the output signal for the new algorithm is larger than that of the algorithm of ASP,AHTP,and AIHT for 0.8dB,1.38dB and 3.13 dB respectively and is less than that of the algorithm of GAP and AL1 for 0.32 dB and 0.6dB when the input signal is with Gaussion noise.The complete reconstruction probability of the new algorithm was greatly improved by adopting the above measures,and the convergence condition for the new algorithm was simplified.
摘要:The aim of this paper is to investigate the influence of rail corrugation on time-frequency characteristics of a train.A spatial coupling model of the vehicle-track system is presented,and dynamic response of the vehicle system under the track short wave and long wave corrugation stimulation is obtained.The Ensemble Empirical Mode Decomposition and Normalized Hilbert Transform are introduced to the vibration signal analysis of vehicle-track coupling system.Then the influence of different wavelength and wave depth of corrugation on time-frequency distribution of bogie response are analyzed.The simulation result indicate that corrugation can induce the frequency modulation of bogie response,the ratio of frequency amplitude to mean frequency can be used as an indicator of rail-wheel impact.
摘要:In situations where the targets cannot be detected in the surveillance region,the estimated performance of the adaptive target birth intensity probability hypothesis density (PHD) filter will get worse because of false or low estimate.To overcome this problem,with unknown target birth intensity,a single measurement PHD (PHD-SM) filter and its sequential Monte Carlo (SMC) method are proposed.First,the undetected targets are compensated through developing the one step virtual measurement set.Afterward,according to the single measurement decomposition technique of PHD,the predication and update equations are derived.Finally,a novel multi-target state estimation method is presented.The simulation results show that,when the detection probability PD is small,PHD-SM filter has higher estimation performance.Moreover,the smaller the detection probability,the more significant advantage of estimation performance for PHD-SM filter.
摘要:In order to solve the over-fitting problem of the traditional pattern recognition approaches under the DNA microarray data with small train samples,a subspace fusion-based evolutionary hypernetwork model is proposed in this paper.With the methods of subspace division,hyperedge coverage,and subspace fusion,the proposed scheme reduces the dependence on the initialization,decreases the fitting error of the data space,and enhances the generalization ability of the evolutionary hypernetwork.The experimental results on four DNA microarray datasets show that the proposed model achieves higher classification accuracy and stronger generalization ability than other compared traditional pattern recognition method.
关键词:pattern recognition;microarray data classification;evolutionary hypernetwork;subspace;over-fitting
摘要:Networks' energy efficiency and frequency efficiency are regarded as current hot topic,which have an significant impact on network performance.However,the in-deep studies about energy efficiency and frequency efficiency problem and mobile Ad hoc networks are absent.More studies only consider networks' energy efficiency and frequency efficiency of single object,while it is very few to combine networks' energy efficiency and frequency efficiency.This paper analyzes energy efficiency and frequency efficiency problem in mobile Ad Hoc networks.Then we discuss the tradeoff between both them.Multi-objective optimization theory is used to build multi-objective optimization model combining networks' energy efficiency and frequency efficiency.The weight sum method is exploited to convert the multi-objective optimization into the single-objective optimization.And we propose a notation of resource efficiency.Finally,we present a routing scheme for resource efficiency optimization in mobile Ad Hoc networks.Simulation results show that the proposed algorithm exhibits better performance.
关键词:mobile ad hoc networks;energy efficiency;multi-objective optimization;frequency efficiency;routing algorithm
摘要:As the traditional clustering algorithm cannot be applied to the cognitive ad hoc network for dynamic channels,a distributed clustering algorithm based on the similarity of channels has been proposed.Firstly the channel similarity between nodes will be calculated and the probability of a node within the cluster will be estimated using an adapted EM algorithm.Then by using minimum cut algorithm in graph theory,the optimal clustering results will be obtained with maximum similarity within a cluster and minimum similarity between clusters.Finally,a coordination mechanism to synchronize the global clustering information has been proposed.Throughout,these processes are evenly distributed,without relying on a common control channel.The simulation results show that the proposed algorithm can change the cluster structure according to the dynamic nature of channels,increase the intra-cluster common channels,and effectively reduce inter-cluster common channels to lower the interference.
关键词:cognitive Ad Hoc network;clustering;channel similarity
摘要:Multi-label classification in network environments is becoming a key area of data mining research as its applications are increasing dramatically.Relational classification models,which predict class labels of linked neighbors according to the ones of the given nodes,have been shown to outperform traditional multi-label classifiers.However,existing relational classification models neither make full use of neighbor information,nor predict the isolated nodes' labels,which are popularly existing in relational networks.In this paper,we present a multi-label relational classifier (MORN) that mines both second-order neighbors for non-isolated nodes and high-order neighbors for isolated nodes.MORN has been conducted on real datasets and it demonstrates that our proposed classifier outperforms existing relational classification models.
摘要:This paper deals with the problem of tracking target in a distributed camera network using the cubature Kalman filter.The square-root cubature Information filter (SCIF) is an extension of the cubature Kalman filter.It is an efficient and robust non-linear filter for multi-sensor data fusion.However,when this algorithm is applied to large-scale networks such as camera networks,the center node may be imposed on severe computational loads if using centralized multi-sensor system.In order to solve this problem,a distributed algorithm based on square-root cubature information filter is presented for large-scale networks.In camera networks,because cameras are arranged in a larger region,the target may appear in the blind zone due to the limited field of view (FOV).This may produce invalid measurements from some cameras.To overcome this problem,this paper proposes a novel square-root cubature information weighted consensus filter (SCIWCF) which reduces the effect of these invalid measurements in consensus algorithm via proper weighting on the information vector and information matrix.The simulation results show that the proposed algorithm can efficiently track the target in camera networks,and is obviously better in terms of its accuracy and stability than the traditional Information filter.
关键词:cubature Kalman filter;information filter;consensus algorithm;distributed camera networks
摘要:A new face recognition algorithm via bag-of-words(BoW) is proposed.In specific,it uses BoW and the global pattern of BoW respectively as the local feature and global feature of face images.Multiple kernel learning is adopted to fuse the local and global features.Extensive experiments were carried out on four face databases,i.e.AR,FERET,CMU PIE and LFW.The results show that our method can effectively solve the small training size problem and is more robust to expression changes,position variations and occlusion.
摘要:Few existing image segmentation methods simultaneously take into account both the distinguishability of different features and the relationship between neighboring pixels.In this paper,a novel image segmentation algorithm is proposed by combining the adaptively weighted features with the Gibbs random field.First,the distinguishability of each component of image features for each land-cover type is defined as a weight parameter,which is determined by the corresponding component of the training samples belonging to the same land-cover type.Second,the initial segmentation is obtained by using the minimum distance classifier,and the spatial correlations of neighboring pixels are modeled by the Gibbs random field.Finally,the label field,which is modeled as the label prior of Gibbs random field,and feature field,which is represented as the normalized weighted distance of weighted features,are combined together to generate the segmentation result.Experimental results demonstrate that the Gibbs random field can effectively describe the spatial relationship,and the adjusted weight can strengthen the distinguishability of the feature component,which can distinguish different land-cover objects accurately.
关键词:image segmentation;Gibbs random field;distinguishability;feature weighting
摘要:To tackle video management problem of semantics gap existing in different aspects,a video semantic description framework based on UCL (Uniform Content Locator) is proposed.The semantic description framework consists of three levels,i.e.,content,control and physical.Video to be semantically managed is divided into different scenes based on spatial-temporal similarities of frames.For every scene,the most optimal reference frame (I-frames) and non-optimal reference frames (non I-frames) are selected based on local texture complexity,background luminance and scene complexity.Content semantic are imbedded into non I-frames while control and physical semantics are imbedded into I-frames.A semantic watermarking algorithm is incorporated into the management to realize the efficient storage and transmission of video content and its video semantics.JM reference model is adopted for experiments to verify the watermarking technique and results show that the method is robust and has little side effect on video quality.
摘要:Under the YCbCr color space mechanism,we propose a coupled total variation model for image colorization by combining the gradient of target grayscale image.Then we design a fast algorithm to solve the proposed model numerically by incorporating the alternating direction method of multipliers(ADMM) algorithm,and give the convergence results of the proposed algorithm.Finally,experimental results are reported,demonstrating that the proposed model can effectively prevent colors from crossing over the edges of the target grayscale image during fast colorization.
关键词:image processing;image colorization;coupled total variation;alternating direction method of multipliers(ADMM) algorithm
摘要:Sinusoidal frequency is one of the basic parameters.The high-accuracy sinusoidal frequency measurement technology has wide applications.But in the aspect of low-frequency sinusoidal frequency measurement,the current sinusoidal frequency measurement technology usually has the defects of low accuracy under the environment of harmonic noise interference.This paper proposes a novel sinusoidal frequency measurement method based on precise calculation of initial phase by means of sequence and deconvolution sequence.The measurement principles of precise initial phase sequence calculation and orthogonal mixing calculation for sequence phase measurement are analyzed.It is pointed out that the mixed-frequency interference is the main cause of the measurement error.By severely inhibiting the interference of mixed-frequency,and by calculating the full phase difference of sequence and deconvolution sequence,accuracy of sinusoidal frequency measurement is improved under the environment of harmonic noise interference.Mathematical calculation,simulation test and the physical experiment results also verify the correctness and reliability of the proposed method.
摘要:To mitigate DME pulse interference over L-DACS1 and overcome the shortage of conventional solutions in time,frequency or second-order statistics domain,an innovative adaptive interference cancellation scheme by the usage of high-order statistics is introduced,namely CE-LMS.CE-LMS models the coexistence of L-DACS1 signal and DME interference as a deterministic signal plus colored Gaussian noises,and employs a cost function based on the three-order cumulant statistics and a variable step adjustment mechanism based on logarithmic function.All this together allow CE-LMS to eliminate the pulse-like DME interference from L-DACS1.Numerical results show that CE-LMS algorithm can achieve a better cancellation ratio and lower bit error rate with a increase in computational complexity,which may help to the deployment of L-DACS1.
关键词:L-DACS1(L-band Digital Aeronautical Communication System Type 1);DME(Distance Measuring Equipment);high-order statistic;interference cancellation;adaptive filtering
摘要:Aiming at the effects of slow convergence speed and large computational loads of nonlinear satellite channel Volterra blind equalization system,a nonlinear satellite channel blind equalization algorithm based on multiwavelet double transformation is proposed.In the proposed algorithm,Volterra equalizer of the nonlinear satellite channel is replaced with Wiener equalizer to reduce the complexity of the equalizer structure.And the inputs of the Wiener equalizer are transformed via using balanced orthogonal multiwavelet transform to reduce the autocorrelation of the inputs.In addition,the decision feedback filter is added to the output terminal of the Wiener equalizer,as well as the inputs to the decision feedback filter are also transformed via using balanced multiwavelet transform to further reduce the their autocorrelation.The simulation results have verified the effectiveness of the proposed algorithm.
摘要:Targeted at on-line microbloggings,on the basis of weighted and dynamic link prediction features,we utilize nonnegative matrix factorization to predict existence and directivity of link from user-based and post-based dimension by employing relationship-based factor to constrain objective function.Experiments on real-world dataset demonstrate the effectiveness of the proposed framework.Further experiments are conducted to understand the importance of features' weights and temporal information in link prediction.
关键词:directed link prediction;nonnegative matrix factorization;features' weights;temporal information;dynamic social networks
摘要:For the roller eccentric signal of upper and lower support rolls is intensive spectrum.In this paper,a new roller eccentric harmonic parameter estimation method is presented,which is the harmonic parameter estimation based on multiple modulation zoom spectrum analysis.Using multiple modulation zoom spectrum analysis algorithm to accurately estimate the frequency and amplitude of eccentricity harmonics;At the same time,combined with the FFT (Fast Fourier Transform) algorithm to estimate phase parameters of the eccentric harmonic.The simulation results show that this method is effective and feasible for eccentric harmonic parameter estimation.
摘要:Code-reuse attacks (CRAs) are difficult to detect and defend,especially on widely used x86 processors.One reason is that lots of unintended but legal instructions exist in x86 binary codes.The unintended instructions make the finding of so called gadgets for CRAs is much easier than that of RISC processors.Previous studies rely on software-only means to tackle the unintended instruction problem,which makes their approaches are either very costly or can only be applied under restricted conditions.In this paper,we propose a hardware/software co-design approach to tackle the unintended instruction problem.The proposed mechanism has little performance impact on the examined SPEC CPU 2006 benchmarks.We also propose using hardware control-flow locking as a complementary technique.By using the two techniques together,an attacker will have little chance to carry out CRAs on x86 processors.
摘要:Malware homology identification is useful for malware authorship attribution,attack scenario restoration,and so on.Current malware homology identification methods still rely on manual analysis,which is inefficient and time-consuming.In order to improve the effectiveness and efficiency,an automatic malware homology identification method is proposed.Based on 7-class calling behaviors,this method constructs a model of calling habits using data mining algorithms.Then it calculates the degree of homology based on Frequent Pattern Outlier Factor.Finally,it chooses the threshold values using k-means clustering algorithm to identify homology.The experimental evaluations on real-world malwares show our method achieves high accuracy (over 99%) and acceptable recall rate.
摘要:The paper proposes the model of network survivability based on the patterns of normal network attack and APT attack.The model is demonstrated by constructing the simulation scenarios of network attack to analyze their performances.The results show that this evaluation model can effectively reflect two parameters: the speeds of network attack propagation and network recovery,and the performance of network survivability of APT attack pattern is lower than that of normal attack pattern.
摘要:Achieving KNN query with traditional cloaking region brings higher communication cost and delay caused by useless points of interest (PoI) returned,a new KNN query method is proposed.Based on Voronoi diagram division of PoIs and hierarchical index quadtree structure,cloaking region can be constructed purposefully.Due to the targeted query request,the communication cost is decreasing compared with traditional cloaking region methods.And injecting fake query requests makes the query content privacy preserving work.We have verified the effectiveness of our proposal by analysis and experiments.
关键词:location privacy;location based service;cloaking region;K nearest neighbor query
摘要:Heterogeneous signcryption,a cryptographic primitive,can simultaneously provide the confidentiality and unforgeability of data transmission between different public key cryptography.The paper gives the definition and security models of CLPKC-to-TPKI heterogeneous signcryption scheme between CLPKC (Certificateless Public Key Cryptography) and TPKI (Traditional Public Key Infrastructure),and presents a construction by using the bilinear pairing.In the random oracle model,based on the assumptions of Computational Diffie-Hellman and modifying Inverse Computational Diffie-Hellman,the scheme is proved to satisfy the confidentiality and unforgeability of the insider security.Moreover,it satisfies the properties of ciphertext anonymity which can efficiently protect the privacies of sender and receiver.Owing to the independence and difference of the system parameters in CLPKC and TPKI,the scheme is more suitable in the practical environments.Furthermore,the analysis of efficiency shows that,comparing to the existing heterogeneous signcryption schemes,the scheme is more efficient,so it is suitable for the requirements of identity hiding and constrained bandwidth.
关键词:heterogeneous signcryption;certificateless public key cryptography;ciphertext anonymity;computational Diffie-Hellman problem (CDH);modification inverse computational Diffie-Hellman problem(mICDH)
摘要:Existing source localization methods mostly assume that the sources are pure near-field sources or pure far-field sources.While in practical localization systems,both far-field and near-field sources may exist simultaneously.To classify far-field and near-field sources,and also to achieve high-precision source localization,a novel mixed far-field and near-field source localization algorithm is proposed in sparse signal reconstruction framework.The algorithm first utilizes anti-diagonal elements of array covariance matrix and reweighted l1-norm penalty to obtain DOA estimation of all sources,then classifies far-field and near-field sources and successively obtains range estimation of near-field sources via one-dimensional search,by exploring the feature that the range parameters of far-field and near-field sources are located in different areas.Theoretically,we analyze the reconstruction performance of the reweighted l1-norm penalty algorithm.The proposed algorithm is not only suitable for dealing with Gaussian signals and non-Gaussian signals,but also without multi-dimensional search and parameter pairing process,and also without knowing the number of sources.Meanwhile,the proposed algorithm can even provide good estimation accuracy.Computer simulation results validate the effectiveness of the proposed algorithm.
关键词:array signal processing;far-field sources;near-field sources;mixed far-field and near-field sources;reweighted l1-norm penalty
摘要:Most of the opportunistic networks routing algorithms based on community detection use an un-weighted network which ignores the degree of intensity of relations between nodes.This paper proposes a routing algorithm based on community detection of weighted network.We improve quick community adaptation (QCA) and make it adapt to the opportunistic networks by using weighted networks.The algorithm calculates link weights by the connection information between nodes in the network.According to the different network environments,we present two weight calculation strategies:normalized weight strategy and non-normalized weight strategy.The algorithm detects the environment around the current node,and then chooses the right strategy.To illustrate the performance of our algorithm,we test the algorithm by using a simulation environment and a real dataset.The results demonstrate that our algorithm gets a reasonable community structure and reduces the overhead ratio and keeps a higher delivery probability.
摘要:With the development of social media,massive reviews are generated by users every day.The extraction of evaluation elements,including evaluation phrase,comment target and opinion holder,is an important pre-task of Chinese opinion mining and sentiment analysis.This paper proposes an efficient method for extracting Chinese evaluation elements based on cascaded model and mainly makes three contributions:(i) to implement and evaluate the method,we construct an original annotated corpus for Chinese evaluation elements of automobile;(ii) we provide specific definition and classification of Chines evaluation phrase;(iii) combing statistic method and rule-based method,we present cascaded strategy for extraction of evaluation phrase and evaluation elements,respectively.According to the experiment results,the proposed method performs well,and effectively improve the recall compared with other rule-based algorithm.Meanwhile it contributes greatly to our subsequent tasks,such as sentiment analysis of social media.
摘要:The hot phrases in the social network text streams can reflect the hidden hot topics and sudden events.This paper proposes a hot phrase mining technology which can support various hot degree measures without word segmentation.We first construct an AC-Trie using the candidate phrases gathered from text streams.Based on such AC-Trie,we record the historical occurrence frequency of phrases on the Trie by scanning the following streams in single-pass.Furthermore,the AC-Trie needs to be reconstructed using the new samples in the text stream because of the evolution of hot phrases.Thus,we start the reconstruction dynamically according to estimating the occurrence frequency of the missed phrases.The experiments on the Sina micro-blog show that our approach is effective (precision of 89%) and efficient (overhead is 2% of naïve approach).
摘要:A lot of news articles in the portal,blog and forums always have their own emotional orientations and sentiment key sentence identification plays an important role in distinguishing emotional orientation of one article,supervising social trends and public sentiment state.The traditional lexicon-based methods totally depended on lexical semantics and did not excavate the implied syntactic structure.So a hybrid method of sentiment key sentence identification based on lexical semantics and syntactic dependency is proposed in this paper.This approach first gets lexical semantics knowledge from emotion lexicon expansion and keywords lexicon construction,and then this paper proposes a novel dependency templates extraction algorithm for syntactic dependency information to build a dependency knowledge base,finally we regard sentiment key sentence identification as a classification task and perform identification through different groups of features.Experimental results on COAE2014 dataset show that this approach notably outperforms other baselines of sentiment key sentence identification on precision and recall.
摘要:Aiming to the privacy-preserving problem for moving-object retrieval services in social network,we propose a granular friend retrieval framework based on over-anonymity,called F-Seeker.Before outsourcing data,we adopt an enhanced anonymity strategy——(k,m,e)-anonymity,which preserving user privacy from the curious retrieval service provider.In the processing of providing services,the service provider employs over-anonymity strategy based on visibility requirements to realize granular data access control.In addition,we encode data using Z-order address and the retrieval efficiency can be improved by pruning.Experimental results show that the proposed strategy can protect user privacy while the computation overhead does not increase greatly.
关键词:over-anonymity;granular search;location-based service (LBS);Voronoi diagram;Z-order space filling curve
摘要:Implementation of file fault tolerance is the key to preventing data loss in cloud,especially the off-site disaster recovery,which can prevent large-scale power outage and natural disaster occurs from losing data.Currently disaster recovery guarantees for multi-cloud-based storage service provider agreement.To solve these problems,we propose a remote disaster recovery capability for cloud data to validate the program——DPBDL,whose theory is to use time delay and the proofs of retrievability to check the off-site data recovery of cloud data.Then,we analysis its security and performance theoretically and practically,and the analysis result shows that its security and be able to accurately determine the cloud data off-site disaster recovery capabilities.
摘要:A novel miniaturized planar monopole antenna for ultra-wideband (UWB) applications with dual band-notched function is proposed in this paper.The proposed antenna consists of an ordinary square radiating patch with a π-shaped slot,a modified W-shaped slit and a pair of L-shaped branches.By adding the pair of L-shaped branches,additional resonances are excited and wider impedance bandwidth can be produced.To generate dual band-notched characteristics,a π-shaped slot in the radiating patch surrounded by a modified W-shaped slit is etched.The antenna provides a wide usable fractional bandwidth of more than 130% (2.77-13.62GHz).In addition,the designed antenna has a small size of 12×18×1.6mm3 while showing the band-rejection performance,covering all the 3.5/5.5-GHz WiMAX,4-GHz C-bands and 5.2/5.8-GHz WLAN.Simulated and measured results show good band-notched functions,stable gain and nearly omnidirectional radiation patterns of the proposed antenna over the entire UWB frequency.
摘要:Due to the absence of information concerning the location and number of scattering objects in inverse scattering problem,it results in nonlinear inverse scattering method calling for high refactoring space dimension and high computational cost.Based on the low computational burden with the linear inverse scattering methods in reconstructing the location and number of the targets,we propose a nonlinear inverse scattering method in this paper,which combined with DSM linear and CSI nonlinear inverse scattering methods.The reconstruction results of DSM are used as the prior information to provide the location and the number of the targets for the CSI method,which reduce the dimension of refactoring space and the amount of computation.Experimental results show the effectiveness of the proposed method.
摘要:The circuit offset often causes integration saturation in the traditional sigma-delta interface of capacitive MEMS accelerometers.To address this problem,a new type of capacitive digital interface circuit used for downhole exploration and oil detection is designed.This paper presents a MEMS-based 5th-order sigma-delta capacitive accelerometer,where the 3rd-order digital loop filter is realized using FPGA.This will reduce the ASIC analog circuit layout design and chip testing difficulties and is easy to optimize the loop filter parameters,which can be used to improve the system stability and optimize the noise performance.The analog-frond-end amplifier (AFE) is realized by using a simple correlated double sampling (CDS),which is one effective method to reduce circuit offset of AFE.According to the Gaussian distribution of AFE output signal,one new type 8-bit instantaneous floating point ADC (IFP ADC) is designed.The IFP ADC is used to convert analog signal of AFE to digital signal to feed the 3rd-order digital loop filter.The whole system provides a significantly low noise floor 53.09ng/rt(Hz) overall a 200 Hz bandwidth.In this work,the AFE amplifier and ADC were successfully fabricated by using XFAB XH018 mixed-signal CMOS process.Furthermore,the sensitivity and noise floor of the AFE amplifier are 0.69V/pF and 3.20 μV/rt(Hz) in open loop measurement,respectively.
摘要:Model-based diagnosis,an intelligent diagnosis theory has been successfully applied in software fault localization with promising results.However,traditional MBD relies on the assumption that components in the system fail dependently which makes the diagnoses with high false positives in software fault localization.In this paper,a component redundancy analysis approach is presented.The approach not only uses Craig interpolant to filter redundant components,but also employs a fact that a branch predicate evaluates to either true or false to filter some branch condition components.Experimental results show that the proposed approach effectively reduces the false positive rates of diagnoses,i.e.,reducing the number of diagnosis by 48.4%,and reducing the number of nodes of hitting set tree generated during diagnosis computation by 47.6%.
摘要:User demographic attributes,such as gender and age,are the core factors to be considered for research and applications in computational psychology,personalized search and social commerce marketing.Automatic user latent attribute inference based on user generated data becomes an emerging research topic.This paper proposes a methed for user gender inference on Microblog by exploiting user content preferences and following behaviour preferences.The experiments on a dataset collected from Sina Weibo that consists of nearly 10000 users demonstrate the effectiveness of user preferences features.Comparing with the traditional language usage features,combining user content preferences and user following preferences features can improve the inference accuracy largely.The user following preferences features are especially effective for inferring the gender of inactive users.
关键词:user latent attribute;user gender inference;user preference modeling;social media
摘要:The intuitionistic fuzzy c-means clustering algorithm cannot discover the non-convex cluster structure.To alleviate this problem,an intuitionistic fuzzy c-means clustering algorithm based on kernelled distance is proposed.By defining the intuitionistic fuzzy Euclid distance,we map the sample to a high-dimension feature space.So the former features can be reflected thoroughly,which is helpful for clustering.Experiments executed on one artificial data sets and one UCI data sets demonstrate the performance of the proposed method.Compared with the five classical cluster algorithms,our method is of obvious effectiveness and superiority.
摘要:The adaptive operator selection method is used to solve the global optimization problem and multi-objective optimization problem of differential evolution algorithm.However,it is difficult to find a way to properly allocate credit for the adaptive operator selection in solving the constrained optimization problem.In order to realize the adaptive strategy selection in differential evolution,we present a combined population based adaptive fitness method to achieve the credit assignment of mutate strategies for constrained optimization problems and use probability matching method to select the mutate strategy adaptively.And we also set the mutation scaling factor and the crossover rate adaptively to improve the success rate of the algorithm.Experimental results show that the algorithm has higher accuracy and convergence speed comparing to CODEA/OED,ATMES,εBBO-dm,COMDE and εDE.We also test and verify the effectiveness of the adaptive method.The algorithm can be used in forecasting,quality control,accounting process,and other scientific and engineering applications.
摘要:In view of some problems,like slow convergence speed and low constringency rate,arising during the process of applying discrete bat algorithms to solve travelling salesman problem,a chaotic hybrid discrete bat algorithm is proposed.The proposed algorithm adopts chaotic initialization strategy to improve the capability of optimization,and the 2-Opt to enhance the capability of local search and to speed up the convergence speed.A large amount of simulations show that the algorithm can achieve their solutions rapidly for some small scale traveling salesman problems,and obtain their solutions in a relatively short time with the error less than 0.4% for large ones.
摘要:The geometric model and analysis model can be uniformly described by NURBS in isogeometric analysis method,so the model transformation process is avoided.However,since the analysis domain of IGA should be composed of tensor-product patches,it is difficult to deal with the issue of complex multiply connected domains.IGA based on trimming technique is constructed by NURBS geometric modeling of underlying surface and trimming curve,and then,directly map a parameter space trimmed by NURBS curve as the complicated holed structure,only one parameter space is sufficient to describe arbitrary complex topology.The advantages and properties of conventional IGA are maintained,and also,the range application of IGA is enlarged.In this paper,IGA based on trimming technique is expanded to the TE waveguide eigenvalue problem.With the solution of cutoff wave number for complicated multiholes structure,the effectiveness and high accuracy of proposed method are demonstrated by corresponding numerical examples.
摘要:The MUSER(Mingantu SpEctral Radioheliograph) is a new generation of solar dedicated radio telescope.Its ultra-wide band dual-circular polarized feed is one of the key technologies.AR (Axis Ratio) is an important parameter for measuring feed efficiency.The traditional AR measurement needs the involvement of the standard antenna.In this paper,the method by measuring correlation is adopted to measure AR of feeds for MUSER.Since MUSER has numbers of antennas and located in harsh environments,according to the demand of astronomical observations,the performance of the antennas should be measured frequently.This method simplifies the process of AR measurement of the interferometric array and can obtain accurate results efficiently.The measuring process are presented in detail in this paper.In addition,the results of AR measurement show the favourable characteristics of the feeds for MUSER.
关键词:synthetic aperture;Mingantu spectral radioheliograph;circular polarization;axis ratio