15 April 2018, Volume 44 Issue 4
-
Select all|
-
Computer Engineering. 2018, 44(4): 1-11. https://doi.org/10.3969/j.issn.1000-3428.2018.04.001
To meet the real-time processing needs of compute-intensive big data applications,H-Storm heterogeneous computing platform TS developed based on Apache Storm.Through the Multi-process Service(MPS) feature,Graphic Process Unit(GPU) resource quantization and distributed calling mechanism are designed the task scheduling strategy of H-Storm heterogeneous clusters is proposed,and the task scheduling algorithm of GPU performance and load and adaptive flow distribution decision mechanism under cooperative computing are realized.Experimental results show that in the case of 512×512 matrix multiplication,the throughput of H-Storm heterogeneous computing platform increases by 54.9 times and the response delay decreases by 77 times compared with that of native Storm.
-
Computer Engineering. 2018, 44(4): 12-16. https://doi.org/10.3969/j.issn.1000-3428.2018.04.002In view of the tide problem of fast developed city public bike system,this paper proposes a station clustering algorithm based on SimRank,which uses the characteristics of public bike.Firstly,the definition of station similarity is proposed based on the relation between stations.Secondly,the SimRank algorithm is introduced to calculate the similarity between stations.Finally,according to the calculated similarity values,the stations are clustered with the idea of maximum similarity priority.Experimental results show that the clustering results by the proposed algorithm have accurate bike flow characteristic and regional characteristic,meanwhile,the members in same cluster have great relevance.
-
Computer Engineering. 2018, 44(4): 17-22,27. https://doi.org/10.3969/j.issn.1000-3428.2018.04.003In the process of distributed processing of all-to-all comparison problem for large data,the existing data allocation strategies think less of the special dependency between the comparison task and the data,which lead to the low storage efficiency and imbalanced task allocation.Aiming at this problem,a Data Allocation Algorithm Based on Graph Covering(DAABGC) is proposed.Firstly,the problem of data allocation for large data is summarized as the problem of graph covering by theoretical analysis.Then,the optimal solution of several graph covering is constructed successfully and the data are allocated according to the special solution.Experimental results show that,compared with the Hadoop-based data allocation strategy,the proposed algorithm ensures that the comparison task has 100% data locality and load balancing between nodes.It also improve storage saving rate and overall computing performance.
-
Computer Engineering. 2018, 44(4): 23-27. https://doi.org/10.3969/j.issn.1000-3428.2018.04.004Aiming at the problems of low operational efficiency and low coverage in the late-night bus routes in China’s big cities,this paper introduces a bus route evaluation model and Dijkstra algorithm model considering the regional equilibrium,and proposes an improved scheme for late-night bus lines in the city.Taking the late-night bus network in Shanghai as an example,the late-night mobile demand display module in urban areas,the late-night bus network evaluation module in urban areas and the new late-night bus network design module are researched to realize the integrated design of evaluation to optimization.By establishing an effective evaluation and optimization system of urban public transportation network,the key technologies of mobile data application in urban public transportation network optimization are discussed.Optimization results show that the improved scheme of late-night bus network basically covers all the demand for late-night mobile and taxi mobile data high,compared with the existing late-night bus lines,the scheme to meets the late-night travel needs in Shanghai.
-
Computer Engineering. 2018, 44(4): 28-34,40. https://doi.org/10.3969/j.issn.1000-3428.2018.04.005Selecting the proper service provider for different users has many important problems.The most important one comes from the heterogeneity and dynamic of user’s network.Therefore,based on the sea service architecture,a measurement and recommendation system oriented online video services is designed and implemented.The system simulates a large number of client nodes to measure and predict the user’s Quality of Experience(QoE) based on the measurement results,thereby providing users with real-time service source recommendations.The system is used to build a video measurement and recommendation system containing 10 video sites,and the data for 9 months are observed and analyzed.Experimental results show that the system can provide service source recommendation to users needs to watch,so that the user can use the system to obtain the best viewing experience under the current network conditions.
-
Computer Engineering. 2018, 44(4): 35-40. https://doi.org/10.3969/j.issn.1000-3428.2018.04.006Aiming at the problem that the Nearest Neighbor Absorption First(NNAF) clustering algorithm is difficult to be applied in the massive data clustering process,an improved algorithm is proposed based on MapReduce.By introducing MapReduce parallel programming framework and using Canopy coarse clustering,it optimizes the calculation process and improves the process of clustering the intersection.Three different data sets are used to compare the K-means algorithm,the improved NNAF clustering algorithm and the NNAF clustering algorithm.Experimental results show that the improved algorithm can guarantee the clustering quality to a certain extent and has higher running speed.It is suitable for clustering analysis of massive data.
-
Computer Engineering. 2018, 44(4): 41-45,65. https://doi.org/10.3969/j.issn.1000-3428.2018.04.007The existing routing algorithms do not consider the real-time transmission status and traffic characteristics of the link,a multi-path routing algorithm based on link real-time status and traffic characteristics is proposed based on the idea of Software Defined Network(SDN) centralized control and whole network control.The algorithm divides the data stream into the big stream and the small stream,and has the characteristics of high demand for the large stream throughput,routing according to the path weight value,and is available because of the big number of small streams and the low complexity of the small stream processing.The path with the largest remaining bandwidth is selected as its routing path.Simulation results show that this algorithm can improve the average link utilization and network throughput of fat-tree data center network compared with Equal Cost Multi-path(ECMP) and Software-defined Hybrid Routing(SHR) mechanism.
-
Computer Engineering. 2018, 44(4): 46-51. https://doi.org/10.3969/j.issn.1000-3428.2018.04.008To reduce the influence of bandwidth for mobile transparent computing system TCOS,this paper proposes a cache cooperation strategy named TC-CCS that combines the local cache and the P2P cache.Local cache is used to store the user’s personal data.P2P cache is divided into two parts,which is used to store the operating system and the starting data of application software.According to the feature of data access in TCOS system,the paper designs and implements a TC-replace replacement algorithm,which can replace the cache flexibly by analysis the access of local and other terminals.Experimental results show that,by using the TC-CCS strategy,the starting speed of operating system and application software can be accelerated,and the start-up time of operating system is reduced about 40%.
-
Computer Engineering. 2018, 44(4): 52-58. https://doi.org/10.3969/j.issn.1000-3428.2018.04.009In business processes,most methods for calculating the distance between business processes ignore the importance of event log.Aiming at this problem,this paper proposes a new method based on adjacency matrix.Firstly,it extracts the event log stored in the various information management system and converts them to simple event logs that contain only the execution traces of activities.Then,it defines the immediate relationship between activities and constructs the activity adjacency matrix based on the immediate relationship of activities.Finally,based on the definition of matrix norm,it defines the distance between business processes.At the same time,it proves that the defined distance between business processes satisfies the distance metric.Experimental results show that the proposed method can distinguish the selection structure and the parallel structure and find the invisible task.It also has higher computational efficiency.
-
Computer Engineering. 2018, 44(4): 59-65. https://doi.org/10.3969/j.issn.1000-3428.2018.04.010The existing voice-identification methods directly apply English voice-based task goals into the fuzzy self-adaptation loop.However,the ability of recognitionis limited,as the methods directly matched the raw recognized phrases with rules.In order to solve this problem,a structural transformation approach is proposed.Firstly,voice-based task goals are analyzed through the syntax and their key components are extracted.Then,semantic-equivalence sets are established by expanding semantic-relevance words.Based on these keyword sets of task goals,structural transformation orienting to fuzzy rules’ pre-component is finally completed.By constructing task-oriented robot system,it is verified the approach has better recognition ability of voice-based task goals.
-
Computer Engineering. 2018, 44(4): 66-73,80. https://doi.org/10.3969/j.issn.1000-3428.2018.04.011Complex routing architecture in traditional chip-to-chip interconnect protocols always result in low operating frequency.Therefore,this paper proposes a multi-layer point-to-point serial interconnect protocol.The protocol employs layered architecture to improve the flexibility and scalability,cyclic redundancy check implemented in hardware to enhance reliability,and serial transmission module to reduce the area penalty and improve efficiency.Compared with PCI-Express and SRIO,experimental results show that the protocol has the features of low area,high bandwidth and low latency.
-
Computer Engineering. 2018, 44(4): 74-80. https://doi.org/10.3969/j.issn.1000-3428.2018.04.012Soft error caused by Single Event Upset(SEU) from particle striking has become a serious problem in microprocessor design.Soft protection against system can reduce the impact of soft error.Techniques to deal with these errors exist,but will increase the system cost.In order to find an appropriate system protection combination,taking the DSP C6701 chip as the research object,this paper designs an optimal method establishes the multi-objective model of the cost and performance,improves the particle iteration formula of the standard particle swarm algorithm,and applies the modified discrete Particle Swarm Optimization(PSO) algorithm to solve the proposed model.Experimental results show that the optimal method can improve the reliability and avoid increasing the cost of the system.
-
Computer Engineering. 2018, 44(4): 81-88. https://doi.org/10.3969/j.issn.1000-3428.2018.04.013In order to improve the flexibility of Cyber-Physical System(CPS),the hierarchical architecture model of CPS is put forward.The essence and characteristics of CPS are described.The process of event driven is described in CPS.The modular of CPS is taken as an individual,and the hierarchical structure of CPS and the architecture are constructed.Based on the event,the semantic of the hierarchical model is constructed.The results of the practical application show that the model has a clear structure and high flexibility,which is easy for application to understand.
-
Computer Engineering. 2018, 44(4): 89-97. https://doi.org/10.3969/j.issn.1000-3428.2018.04.014Aiming at the problems of low delivery rate and high latency in Delay Tolerant Network(DTN),a new routing algorithm based on historical encounter information and location information is proposed to select the next routing hop.Using the historical encounter information between nodes to filter the most times’ node which is encountered with destination node for message copy.Further,the location information is used to calculate the moving direction of neighbor nodes,copy message to a pair of nodes whose direction angle is larger.Simulation results show that in the case of short TTL and insufficient node buffer,the proposed algorithm can achieve 50% lower average latency and higher delivery ratio compared with epidemic algorithm which can copy message at once when meets any node.
-
Computer Engineering. 2018, 44(4): 98-102. https://doi.org/10.3969/j.issn.1000-3428.2018.04.015In order to allocate resources reasonably,for many Device-to-Device(D2D) users exist in the wireless network,this paper proposes a resource allocation algorithm based on stable matching.The algorithm guarantees the Quality of Service(QoS) of cellular users,D2D users can reuse the Resource Block(RB) of the network dynamically,so as to improve the utilization of spectrum.Simulation results show that compared with the static matching algorithm,the given algorithm is closed to the optimal scheme and reduces the computational complexity at the same time.
-
Computer Engineering. 2018, 44(4): 103-107,114. https://doi.org/10.3969/j.issn.1000-3428.2018.04.016In order to improve the effect of data recovery after data loss in Wireless Sensor Network(WSN),a new data recovery algorithm is proposed.The joint sparse decomposition method is used to fuse multiple parameters and extract common components.The influence of initial sparsity and step size on selection is also reduced by using signal block method.Experimental results show that compared with Stagewise Orthogonal Macthing Pursuit(StOMP) algorith and Sparsity Adaptive Matching Pursuit(SAMP) algorithm and so on,the algorithm can reduce the error caused by restoration,improve the accuracy of reconstructed signal,and shorten the time needed for computation effectively.
-
Computer Engineering. 2018, 44(4): 108-114. https://doi.org/10.3969/j.issn.1000-3428.2018.04.017In traditional network environment,network performance and energy consumption can not be taken into account.In order to solve this problem,a new designed network evaluation algorithm is proposed in this paper.In Software Defined Network(SDN),a heuristic algorithm and flow table mechanism are proposed to dynamically adjust network performance and energy consumption.Experimental results show that the algorithm can achieve dynamic adjustment of network performance and energy consumption compared to traditional network only focusing on network performance or energy consumption.The algorithm achieves simple implementation and high link utilization.
-
Computer Engineering. 2018, 44(4): 115-119,128. https://doi.org/10.3969/j.issn.1000-3428.2018.04.018When the data is diverted,considering the diversity of the application data business,allocating the traffic flow according to the characteristics of the transmission business,and analyzing the characteristics of the traffic diversion of single user parallel multi network transmission in the heterogeneous wireless network,a multi-business shunt strategy is proposed.It introduces the concept of economics to establish triage strategy based on utility function,and combines the different types service to consider two factors of consumption and income.Then it constructs the utility function for solving optimization problems,and proves the result which is the global optimal solution.Simulation results show that the proposed strategy can reduce delay and cost effectively compared with the load balancing algorithm.
-
Computer Engineering. 2018, 44(4): 120-128. https://doi.org/10.3969/j.issn.1000-3428.2018.04.019In order to alleviate the ripple effect and improve network throughput,this paper produces a distributed channel assignment scheme with anti-ripple effect based on evolutionary game and promoted fruit-flies optimal algorithm.The genetic mechanism of ripple effect is analyzed,a stable equilibrium game model of terminal node cluster evolution is proposed,and the evaluation of ripple effect rate in the state of stable equilibrium is done to achieve the optimization of channel assignment in single collision domain.Experimental results demonstrate that the proposed scheme is able to alleviate the ripple effect and make the rate of ripple effect decrease from 28% to 8% on average,when channel interference threshold is 0.6 with a stable evolutionary equilibrium state.Compared with C-HYA algorithm,EP-OCA algorithm,RBA algorithm,CDACA algorithm,it has a more stable network throughput.
-
Computer Engineering. 2018, 44(4): 129-134. https://doi.org/10.3969/j.issn.1000-3428.2018.04.020In order to solve the problem as difficult to achieve orthogonalization reception,and the transmission sub channel frequency aliasing serious transmission sub channel frequency aliasing,as well as strong interference between nodes in the process of super density transmission in the current 5G network signal transmission algorithm,a 5G network signal transmission algorithm which can improve the frequency drift problem is proposed in this paper.The super density circular ring sort structure is constructed based on the frequency drift characteristic frequency of the transmission subchannel and the frequency aliasing between the base stations in the cell to orthogonally sort receiving nodes in the superdensity state for reducing the frequency interference between channel-signal,and improving the accuracy of data transmission.The optimization method of azimuth precision is designed based on the difference mechanism and reduce the occurrence of data error rate.Simulation results show that this method has higher data throughput and data transmission quality,and lower transmission error rate compared with the current chaotic frequency drift 5G commonly used in the field of eliminating transmission optimization mechanism high precision adjustment of data transmission mechanism.
-
Computer Engineering. 2018, 44(4): 135-139,144. https://doi.org/10.3969/j.issn.1000-3428.2018.04.021
In the modern data center network,the partition aggregation transmission pattern easily leads to the throughput collapse on the bottleneck link,resulting in TCP Incast phenomenon.This paper proposes a random backoff method to reduce the concurrency degree of TCP flow burst transmission.By calculating the optimal random backoff time,the instantaneous congestion degree of bottleneck link is controlled and the TCP Incast problem is solved effectively.And it models and analyzes in theory the time-out probability and throughput when all TCP flows retreat randomly in the random backoff time interval.Experimental results show that the method can effectively avoid the transmission timeout event.The TCP flow concurrency and network throughput are increased by 2 times and 80 times respectively.
-
Computer Engineering. 2018, 44(4): 140-144. https://doi.org/10.3969/j.issn.1000-3428.2018.04.022Aiming at the anti-jamming and interference suppression in wireless communication,this paper proposes a communication anti-jamming method in polarization domain.It uses oblique projection polarization filter in the signal polarization domain to overcome the shortcomings of the traditional anti-jamming technology in frequency domain.Through the research on the characteristics of oblique projection filters,a variety of interference suppression and anti-jamming techniques are designed in the signal polarization domain,including the optimization of polarized transmit signals,the augmented oblique projection filter of nonlinear vector transform,and the statistics of received signals,to improve the design of oblique projection filter.Simulation results show that,compared with the traditional anti-jamming method with oblique projection filter,the proposed method has better anti-jamming performance and interference suppression performance.
-
Computer Engineering. 2018, 44(4): 145-148,153. https://doi.org/10.3969/j.issn.1000-3428.2018.04.023Applications perform a number of complex tasks in modern Data Center Network(DCN).Small tasks and large tasks share bandwidth fairly is not conducive to reducing the average task completion time.Therefore,this paper presents a congestion control protocol for task size perception.The transmission rate is adjusted according to the task size factor and the trailing factor.Experimental results show that,compared with DCTCP,L2DCT,Baraat protocols,this method can reduce the average time of task application in the network.
-
Computer Engineering. 2018, 44(4): 149-153. https://doi.org/10.3969/j.issn.1000-3428.2018.04.024In order to adapt to the real-time network traffic demand,and find the network fault in time,based on classic online distributed traffic engineering protocol TeXCP,this paper proposes Multi-path Traffic Engineering(MTE) protocol.The protocol consists of the link cost collection module and link transmission probability adjustment module,which corresponds path detection module and load balance module of TeXCP respectively.It adopts the non-additive path maximum utilization rate as the path cost.The stability mechanism of TeXCP is modified and simplified.The results of theory and simulation prove that MTE can converge to the equilibrium state of network traffic,and the stability is better.
-
Computer Engineering. 2018, 44(4): 154-160. https://doi.org/10.3969/j.issn.1000-3428.2018.04.025The SQL Injection Attacks(SQLIAs) is a dangerous and effective Web-based attacks way.According to the characteristics of SQLIAs,that is,any form of attacks will eventually change the logical structure of SQL statement,a method of dynamic detection and defense for the SQLIAs based on Aspect-Oriented Programming(AOP) and the analysis of SQL statement structure is proposed,which can defend against SQLIAs from the root of the attack.Firstly,obtain the location of SQLIAs,signature,and the static SQL statement model automatically by using code Analyzer.Then,capture the executed SQL statement dynamically by using AOP during the process of program execution.Finally,judge whether there is a SQLIAs by making the information obtained by static analysis compare with the dynamic information.The effectiveness of the method is verified by a simple user login function,and the experimental results show that the proposed method can detect and defend the SQLIAs effectively.
-
Computer Engineering. 2018, 44(4): 161-167. https://doi.org/10.3969/j.issn.1000-3428.2018.04.026According to the difference between the network phenomena caused by different jamming attacks,a jamming detection algorithm based on symptom-attack association graph is proposed,which consists of two phases,i.e.learning and detection phases.At the learning phase,by learning from the various samples collected from both jamming and jamming-free scenarios,the nodes divide the network phenomenon based on the changes of various measures and establish a symptom-attack association graph.Then,at the detection phase,each node judges whether it is subjected to jamming attack and attack type on the basis of their own association graph according to their own observations,and evaluates the accuracy of the detection result by matching degree.The simulation results on NS3 validate that the proposed algorithm can efficiently detect and classify the typical jamming attacks,such as reactive,random and constant jamming attacks.
-
Computer Engineering. 2018, 44(4): 168-173,180. https://doi.org/10.3969/j.issn.1000-3428.2018.04.027Fuzzing technology and dynamic symbol execution technology are used to discover program exceptions as the test termination condition without further assessing the threat severity of the program exception.This paper describes three main methods for Crash exploitability analysis.Based on the binary instrumentation platform Pin,this paper proposes a Crash exploitability automated analysis framework based on dynamic taint analysis technique.Experimental results show that the framework can accurately and effectively determine the exploitability extent of Crash.
-
Computer Engineering. 2018, 44(4): 174-180. https://doi.org/10.3969/j.issn.1000-3428.2018.04.028
Tight security proofs need shorter security parameters and better efficiency.Therefor,a new Identity-based Signature(IBS) scheme named IDSSTR is proposed,which has a security specification for Computational Diffie-Hellman(CDH) problems and it is also naturally efficient on-line,no additional conditions is needed for the off-line stage and the verification process is unchanged.In order to shorten the total length of the signed message,a modified version of the IDSSTR with message recovery is goven.Analysis results show that,the difficulty of CDH problem is widely considered to be closely related to the discrete logarithm problem,therefor,the proposed signature scheme provides security assurance for such difficult problems.
-
Computer Engineering. 2018, 44(4): 181-186,192. https://doi.org/10.3969/j.issn.1000-3428.2018.04.029The Correlation Electromagnetic Analysis(CEMA) attacks technology is widely used in side channel analysis attacks for cryptographic chips,but the signal collected is easily affected by external noise and other factors, resulting in a large amount of work.To solve the problem,an improved attack scheme is proposed,which use the PA 303N/BNC preamplifier to enlarge the collected electromagnetic information,and then implement the correlation analysis to improve its attack efficiency.The Advanced Encryption Standard(AES) algorithm implemented on the smart card verifies the proposed scheme,result shows that comparing with the traditional CEMA attack,the proposed scheme only needs less electromagnetic curves to reconstruct the key,the efficiency is higher,and the practicability is stronger.
-
Computer Engineering. 2018, 44(4): 187-192. https://doi.org/10.3969/j.issn.1000-3428.2018.04.030
For the presence of temporary,intermittent,and permanent errors,the processor capturing and executing an incorrect instruction will cause a control flow error to occur.This paper proposes a pure Error Detection of Software Signature(EDSS) algorithm based on table-driven with the study of Controlled Flow Detection by Software Signature(CFDSS) algorithm.It constructs two-dimensional table CFID to store the information of the control flow graph,and the illegal instruction jump is detected by comparing the signatures in the basic block with the signature stored in the CFID table.The EDSS algorithm can detect such errors successfully for the illegal branch jumps of the sharing-branch-fan-nodes that cannot be effectively detected by CFDSS algorithm.Experimental results show that the average error detection coverage of EDSS algorithm is 1.3% higher than that of CFDSS algorithm,and the average error detection rate of the sharing-branch-fan-nodes is about 1.9% higher than that of CFDSS algorithm.
-
Computer Engineering. 2018, 44(4): 193-198. https://doi.org/10.3969/j.issn.1000-3428.2018.04.031Mimic defense structures the system architecture with dynamic heterogeneous redundancy to damage the attack chain.In order to increase the efficiency of arbitration and further improve the overall executive ability of the system,according to the arbitration of mimic defense system,this paper proposes a competitive arbitration model based on majority consensus voting strategy,to increase the execution redundancy without changing the arbitration redundancy.Experimental results indicate that compared with Triplex mimicry system,the model can effectively compensate for the performance loss of a mimic system.
-
Computer Engineering. 2018, 44(4): 199-205. https://doi.org/10.3969/j.issn.1000-3428.2018.04.032For sensitive data protection and user security access issues in electric power enterprise,this paper proposes an access control scheme for user of electrified wire netting.When the user wants to query the data,he must interact with the network server in the power system,and obtain the required data from the internal network server.In this process,the malicious user may fake the legitimate user to make a request and obtains the data illegally.Users are authorized and controlled to access the data through building attribute authorities and using attribute-based encryption to prevent the malicious user from threating the power grid system.The scheme also encrypts the sensitive data and ensures the confidentiality of sensitive data.The results of analysis and evaluation show that the proposed scheme has better security and lower computing cost.
-
Computer Engineering. 2018, 44(4): 206-211. https://doi.org/10.3969/j.issn.1000-3428.2018.04.033In the environment of home sharing,the user’s identity privacy may be leaked in the process of obtaining the file.To solve this problem,a fine-grained file sharing scheme based on attribute encryption is proposed.According to the needs of users,the linear secret sharing scheme is used to make access strategy,then the attribute based encryption algorithm is used to get the attribute based encryption ciphertext.Finally,after verifying whether the user property meets the access policy,the shared file by the property decryption algorithm is obtained.Experimental results show that,when the number of attributes is less than 100,the encryption and decryption efficiency of the proposed scheme can meet the user’s actual needs.
-
Computer Engineering. 2018, 44(4): 212-217. https://doi.org/10.3969/j.issn.1000-3428.2018.04.034The key agreement protocol based on standard lattice has longer key and higher ciphertext expansion rate,and the way of lattice representation needs larger space,while the ideal lattice has shorter key length and higher running efficiency.Therefore,combined with Ring Learning with Error(RLWE) problem,an anonymous password authentication key agreement protocol based on the ideal lattice is proposed.Users use low entropy passwords to authenticate and share session keys through servers,so as to avoid users’ long-term key security being threatened in the process of identity authentication.The analysis results show that,compared with the traditional 2PAKE and 3PAKE protocols,the proposed protocol has higher efficiency and shorter key length,which can resist quantum attacks and is suitable for large-scale network communication.
-
Computer Engineering. 2018, 44(4): 218-224. https://doi.org/10.3969/j.issn.1000-3428.2018.04.035A new recommendation method is proposed based on time and space perception.Considering the user's social relationship,the random walk algorithm is used to get the recommendation activity by constructing the user,location,time,activity,and the information network of friends.Experimental results show that this method can get better results compared with the popular activity recommendation method and the social relation activity recommendation method.
-
Computer Engineering. 2018, 44(4): 225-230. https://doi.org/10.3969/j.issn.1000-3428.2018.04.036Currently,most cross-domain recommendation models have two problems:the prediction error becomes large when the user fails to add a tag to the specified item and when building the model,the user’s history label is not taken into account,which causes the decreasing of recommendation result’s accuracy.In order to solve these two problems,a cross-domain recommendation model is constructed,named TagSVD++,which is based on the SVD++ model and fused with the tag recommendation.This model not only inherits the characteristics of SVD++ model’s using of score data to predict,but also adds the user and the project label information.It uses the tag’s frequency to reflect user preferences and characteristics,and introduces the popular penalty factor to prevent hot tags and popular items from bringing interference to recommendation and predictions.Comparative tests on cross-domain data sets for data simulation which is related to real movie and book website are conducted.Experimental results show that TagSVD++ model can improve the accuracy of cross-domain recommendations effectively.
-
Computer Engineering. 2018, 44(4): 231-235. https://doi.org/10.3969/j.issn.1000-3428.2018.04.037Aiming at the problem of the flexibility and complexity of machines to understand natural language and Tibetan sentences,in view of the different expressions of the same semantic sentences in Tibetan language,this paper proposes a Tibetan paraphrases sentence generation method.Through the parsing of the sentence structure of Tibetan and the internal chunks of sentences,it uses permutation recursive algorithm to generate paraphrases sentence.Experimental results show that different from other languages and Tibetan chunks,the number of chunks in a sentence can generate one or more or even thousands of complex sentences with the same semantic meanings by the proposed method,and the accuracy of automatic generation of Tibetan paraphrases sentences reaches 93.4%.It can be applied to Tibetan-Chinese machine translation,machine translation evaluation,Tibetan QA system and other research fields.
-
Computer Engineering. 2018, 44(4): 236-242,247. https://doi.org/10.3969/j.issn.1000-3428.2018.04.038To accurately predict Click-Through Rate(CTR) and make reasonable use of them for advertisement recommendation,based on tag recommendation technology and Collaborative Filtering(CF) methods,a new hybrid advertising recommendation algorithm is proposed.The advertisement key words are introduced into the similarity calculation of Query pages as labels,the sparsity of the similarity matrix is reduced by using Query page weighting comprehensive similarity measure method,and a search advertisement interest model based on the advertisement key words is given.It uses the Top-N policy to reduce the size of the nearest-neighbor candidate sets and filters out the recommended ads based on the predicted CTR,and adjusts Query weighting comprehensive similarity measure parameters and verifies the scalability of the algorithm by experiment.The experimental results on KDDCUP2012 dataset show that compared with traditional collaborative filtering algorithms,tag-based recommendation algorithms and tag-based and project-based recommendation algorithms,the tag-based collaborative filtering advertising recommendation algorithm has better scalability and better recommended quality.
-
Computer Engineering. 2018, 44(4): 243-247. https://doi.org/10.3969/j.issn.1000-3428.2018.04.039
In order to improve the classification accuracy of convolution neural network,an improved Stacking algorithm combining multiple convolution neural networks is proposed.The convolution neural network is used as the base classifier to classify the data,and the new sample is then classified by the meta-classifier.In order to reduce the dimension and correlation of the input data in meta-classifier,the dimension of output data of base classifier are reducted by Principal Component Analysis(PCA).Experimental results on classification accuracy of data sets show that,compared with traditional Stacking,average posteriori probability based algorithm and class voting based algorithm,the classification accuracy of the proposed algorithm is higher and more stable in similar networks and different types of networks.
-
Computer Engineering. 2018, 44(4): 248-252,261. https://doi.org/10.3969/j.issn.1000-3428.2018.04.040Aiming at the problem of dimensionality curse and unbalance distribution in high dimensional unbalance data,an improved k-Nearest Neighbor(kNN) classification algorithm named HWNN is proposed.The k-occur of samples are regarded as the support for each class at the time of prediction,on this way,the potential negative impact of hubs on kNN classification in high dimensional data is reduced.In order to improve the prediction accuracy for a small number of samples,the distribution ratio of minority groups in the occurrence of all samples k-occur is increased by class weighting.Experimental results on 16 imbalanced UCI datasets show that,the classification results of the proposed algorithm in high-dimensional imbalanced data are better than those of several typical kNN methods,the advantage is also obvious in the unbalanced data of the ordinary dimension.
-
Computer Engineering. 2018, 44(4): 253-261. https://doi.org/10.3969/j.issn.1000-3428.2018.04.041Compared with the general mathematical formula,linear algebra formula has more complex structure and richer syntax and semantics,so it is more difficult to achieve its retrieval.Through the analysis of linear algebraic retrieval,several matching modes of linear algebraic formula retrieval are summarized according to user requirements,and aiming at different matching modes,the hesitation fuzzy membership degree is defined from two aspects of the whole and the part by using semantic hesitation fuzzy set and structure,symbols,grammar,semantic of linear algebraic formulas’characteristics,and the similarity function of the linear algebraic expression is established to realize the sorting of the retrieval results of the linear algebraic formula based on hesitation fuzzy similarity.The experimental results of 6 352 formulas show that the method can realize the ordered output of the result data of the linear algebraic retrieval system.
-
Computer Engineering. 2018, 44(4): 262-267,286. https://doi.org/10.3969/j.issn.1000-3428.2018.04.042Regional distribution is an important feature of moving target,which can be used for target trajectory classification.Previous methods assumed that segments of trajectories distribute as rectangles or Gaussian components,thus the accuracy of trajectory classification is significantly limited.To solve the problem,a novel method that takes use of kernel density estimation and maximum likelihood principle is proposed.It has no assumption on the data,so the problem of trajectory classification accuracy limited by model discomfort is solved.The experimental results show that compared with the existing Minimum Description Length(MDL) partition and Gaussian Mixture Model(GMM),the proposed method is insensitive to parameters,training time is shortened,and track classification accuracy is increased by 5%~15%.
-
Computer Engineering. 2018, 44(4): 268-273,293. https://doi.org/10.3969/j.issn.1000-3428.2018.04.043In order to improve the classification efficiency of unbalanced data sets,this paper proposes a classification model.The sample sampling and classification algorithm are optimized.A new sample set is generated by cyclic sampling of the few samples of the decision boundary,combined with the small sample sets synthesized outside the boundary of the decision-making,then the importance of the sample is improved.Aiming at the problem of hyperplane offset in classification of imbalanced data sets by traditional ε-Support Vector Machine(ε-SVM),the positive and negative penalty coefficients and the mixed kernel function are introduced.The objective entropy value method is used to select the penalty coefficients and the performance of the classification algorithm is improved.Experimental results show that compared with the standard SVM algorithm,the classification is better in the classification of imbalanced data sets,the average F-measure value is increased by 18.1%,and the better classification results are achieved.
-
Computer Engineering. 2018, 44(4): 274-280. https://doi.org/10.3969/j.issn.1000-3428.2018.04.044As a typical texture unit of forest vegetation in remote sensing image,the canopy has prominent structural texture features,but the current segmentation methods make use of this texture insufficiently.Aiming at this problem,based on visual attention mechanism,a kind of forest vegetation remote sensing image texture segmentation method is proposed in this paper.The shape and structure of the canopy in the remote sensing image are used as the visual attention target,and the canopy texture is enhanced by the texture filtering method.The multi-scale canopy Salient Image Disk(SID) model is used to mark canopy and by taking the multi-scale canopy SID as seeds,improved region growing method is used to segment the area of forest vegetation.Experimental results show that the proposed method can accurately mark most typical canopies and improve the segmentation accurate of the forest vegetation.
-
Computer Engineering. 2018, 44(4): 281-286. https://doi.org/10.3969/j.issn.1000-3428.2018.04.045Aiming at the requirement of robustness and real-time running of facial key point tracking in video sequence,this paper proposes a simple and effective real-time tracking method of facial key points.The optical flow method is used to track a number of significant points and select a better initial shape for the next frame.The face frame of the next frame is estimated according to the face shape of the current frame to reduce the dependence on the face detector.To prevent errors accumulation,the face detector restart mechanism is joined.Experimental results on a 300-VW dataset show that this method can achieve robust tracking of 68 face key points with a speed of 30+f/s,which can be used for most face real-time applications.
-
Computer Engineering. 2018, 44(4): 287-293. https://doi.org/10.3969/j.issn.1000-3428.2018.04.046Considering the problem that complex noise interference exists in the practical pavement crack detection,by using the visual difference between cracks and background pixels,an algorithm for automatic detection and recognition of pavement crack with visual attention mechanism is proposed.Firstly,pre-processing methods of grayscale correction and anisotropic diffusion filtering are adopted to remove noises.Then,the salient values are calculated separately using global and local grayscale contrast to generate a comprehensive saliency map,where cracks are roughly positioned on.Finally,according to the shape characteristics of cracks,the shape analysis method is used to remove noises and extract targets to realize the accurate location of cracks.Experimental results show that the proposed algorithm can detect cracks quickly and efficiently.It has higher detection accuracy compared with traditional crack detection algorithms based on threshold segmentation,edge detection or wavelet transform.
-
Computer Engineering. 2018, 44(4): 294-298,304. https://doi.org/10.3969/j.issn.1000-3428.2018.04.047In high-resolution optical remote sensing image change detection,most of the object-oriented method can only use simple features combination to get the object features,which cannot implement design and characteristic extraction for high-level features.Aiming at this problem,a method based on Deep Belief Network(DBN) and object fusion is proposed in this paper.The change detection is transformed into dichotomous problem,and image pixel is used as the classification unit.Multi-scale image feature learning and classification methods are designed in feature learning and classification stage to make full use of contextual information of image targets.On this basis,a fusion method based on object classification is proposed to fuse the classification results classified by DBN,so as to reduce the effects of local noise.Experiments are conducted on the dataset acquired by QuickBird,and the results show that the proposed method can improve the accuracy of image change detection.
-
Computer Engineering. 2018, 44(4): 299-304. https://doi.org/10.3969/j.issn.1000-3428.2018.04.048The existing three-dimensional point cloud model segmentation algorithms cannot segment large components and small components at the same time.Aiming at this problem,in this paper,a segmentation method is proposed based on the significance and weak convexity.Firstly,the point cloud model is segmented into weak bumps by spectral clustering method.On this basis,the significance test is used for extracting smaller projections and the weak bumps with small area but the edge feature points are obvious,to solve the problem of under-segmentation.Finally,through mutual visibility and volume similarity,the region is merged to solve the problem of over-segmentation.Experimental results show that the segmentation results of the proposed method are better than those of unsupervised methods such as Heterogeneous,Constraint Planar.
-
Computer Engineering. 2018, 44(4): 305-309,316. https://doi.org/10.3969/j.issn.1000-3428.2018.04.049The similarity measure method of traditional feature matching is mostly assumed to be a histogram with a single structure.Structural Similarity(SSIM) measures perform well in the field of image quality evaluation,but their robustness is poor.Aiming at this problem,a method to measure the structural similarity is proposed.The third-order tensor structure of feature array is extracted from the spatial grid,the feature information is enriched,the SSIM metric is modified by summation,weights are introduced,and the tensor feature structure is used for the similarity measure to establish high robust similarity measurement method.The dot product form of explicit feature maping is provided to speed up the similarity measure.Experimental results for key-point matching and image retrieval show that,compared with the traditional methods,the proposed method has better matching effect with more robustness and faster computation speed.
-
Computer Engineering. 2018, 44(4): 310-316. https://doi.org/10.3969/j.issn.1000-3428.2018.04.050In image classification tasks,classification models trained from the source domain often do not work well with the image data of the relevant target areas due to changes in image background,lighting,shooting angles,and the like.Therefore,this paper proposes a migration learning method based on deep convolution neural network——Sparse Discriminating Transfer Model(SDTM).The method optimizes the classification function by adaptively learning the diserimination feature distribution of the target area.At the same time combing with the characteristics of preprocessed methods combined to obtain a better complementarity.Experimental results show that SDTM achieves better classification performance on the two standard cross-domain classification datasets of Office-Caltech and Office-31 compared with the existing datum and depth migration methods.
-
Computer Engineering. 2018, 44(4): 317-321. https://doi.org/10.3969/j.issn.1000-3428.2018.04.051When using multiple feature functions to mark different regions for image segmentation,it is necessary to find extreme values for multiple functions,resulting in the increase in computation.Aiming at this problem,a method of multi-level set marking by one function is designed,and the segmentation of different regions is realized by solving the extremum problem of one marking function.It summarizes law of regional marking function,obtains expressions of multiple image segmentation model,and combines it with the Variational Level Set Method(VLSM).At the same time,it accelerates the solution of the energy functional extremum problem by the Alternating Direction Method of Multipliers(ADMM) acceleration algorithm.Experimental results show that the proposed method can realize the segmentation of multiple regions and ensure the robustness and efficiency of segmentation results.
2024
Volume 50 No 4
Volume 50 No 4
(Started in 1975, Monthly)
Sponsored by：
East-China Institute of Computer Technology Shanghai Computer Society
Edited & Published by：
Editorial Board of Computer Engineering
Chief Editor：Jiang Bo
Executive Chief Editor:
Hu Bing
Tel：021-67092217
E-mail: ecice06@ecict.com.cn
CN: 31-1289/TP
ISSN: 1000-3428
Sponsored by：
East-China Institute of Computer Technology Shanghai Computer Society
Edited & Published by：
Editorial Board of Computer Engineering
Chief Editor：Jiang Bo
Executive Chief Editor:
Hu Bing
Tel：021-67092217
E-mail: ecice06@ecict.com.cn
CN: 31-1289/TP
ISSN: 1000-3428