• Volume 27,Issue 8,2018 Table of Contents
    Select All
    Display Type: |
    • Computing Model of Color Constancy Perception Based on Genetic Neural Network

      2018, 27(8):1-9. DOI: 10.15888/j.cnki.csa.006471

      Abstract (1463) HTML (637) PDF 1.79 M (1780) Comment (0) Favorites

      Abstract:In the field of machine vision, color constancy is an important factor in achieving computer vision color correction and maintaining the machine stability to color recognition. By means of the psychophysics, the model gains the color perception data obtained by the human eyes perception, and puts it into the neural network for sample training, then optimizes the connection weights and thresholds of the BP neural network using the genetic algorithm. The color constant perception model is applied to the image color correction, and the correction results are evaluated in terms of the subjective and objective measures, the results show that the established algorithm has high precision and better efficiency, low complexity and less error than the classical algorithm, the color representation of images is more consistent with human perception.

    • >Survey
    • Research Progress and Trend of Text Classification for LDA Topic Model

      2018, 27(8):10-18. DOI: 10.15888/j.cnki.csa.006456

      Abstract (2290) HTML (3968) PDF 1.04 M (3670) Comment (0) Favorites

      Abstract:Text classification is an important research direction in the field of natural language processing. It is found that the research and analysis of text classification can help to classify and manage the information effectively and provide strong support for the application of natural language processing. The existing research has made some achievements at the theoretical and methodological level. Nevertheless, the text classification research involves many aspects such as content, domain, and technology, while the research of each subject is complicated. Therefore, there are many defects and shortcomings, which need further systematic and in-depth research. In this paper, we discuss the related theories of text categorization and Latent Dirichlet Allocation (LDA) topic model for the research of text categorization. Then, we analyze the research status of text classification for LDA topic model from three aspects:technology, method, and application. Some problems and research strategies are presented as well. Finally, future trends of text categorization are summarized.

    • Camera Planning Based on Director Repository in Mobile 3D Animation

      2018, 27(8):19-27. DOI: 10.15888/j.cnki.csa.006487

      Abstract (1401) HTML (601) PDF 1.43 M (1679) Comment (0) Favorites

      Abstract:Relying on the whole process of computer-aided animation automatic generation technology, the ultimate aim of mobile 3D animation automatic generation system is to realize automatic animation of text message content. Camera planning, the direct approach to describe the theme of text message, is the important task of the mobile phone 3D animation automatic generation system, and has established complete qualitative and quantitative planning process. On the basis of the existing relevant theories, this study sums up the common rules in the field of photography, constructs the repository of the camera director by using the semantic web ontology language, accomplishes the two levels rule derivation of the camera's qualitative planning based on the director's repository, and ultimately implements the automatic generation of the camera planning in the mobile 3D animation automatic generation system.

    • Real-Time Detection and Positioning of Moving Target Based on Deep Learning

      2018, 27(8):28-34. DOI: 10.15888/j.cnki.csa.006525

      Abstract (2771) HTML (5221) PDF 1.25 M (2580) Comment (0) Favorites

      Abstract:Aiming at the issues of real-time detection and positioning of movement target, a method of deep learning is proposed. The Single Shot multibox Detector (SSD) detection method is used under Caffe framework, the VGG16 model is used as the basic network model, and the additional feature convolution layers are used to extract the multi-scale convolution features. Then the experimental data set is iteratively trained to get the motion target detection model. The moving objects are detected using the trained model and then positioned through binocular vision positioning method. The experiment results show that the system can reach 40 fps. In the 10 m×10 m scene, the average positioning error is within 6 cm. The system has sound performance both in speed and precision, which provides an effective solution for the real-time detection and positioning of human motion in large and medium-sized scenes.

    • Face Recognition Method Based on Quantized Residual Convolutional Neural Networks

      2018, 27(8):35-41. DOI: 10.15888/j.cnki.csa.006491

      Abstract (2358) HTML (1165) PDF 1.04 M (1970) Comment (0) Favorites

      Abstract:Very deep convolutional neural networks based on residual learning have achieved higher accuracy than other methods for large scale face recognition problem. But the massive floating-point parameters existing in the models need to occupy extensive computational and memory resources, which cannot be satisfied with the demand of occasions with limited resources. Aimed at the solution of this issue, a very deep residual neural network based on network model parameters quantization was designed in this study. In detail, based on the model Face-ResNet, the network was added with batch normalization layers and dropout layers, and also its total layers were deepened. Applying binary quantization to parameters of the designed network models, it can compress the model size substantially and improve computational efficiency with little loss of model recognition accuracy. Both theoretical analysis and experiments prove the effectiveness of the designed method.

    • Analysis of Consumer Sentiment Polarity Based on Chinese Online Review of Catering Industry

      2018, 27(8):42-48. DOI: 10.15888/j.cnki.csa.006488

      Abstract (1380) HTML (1115) PDF 1.01 M (2173) Comment (0) Favorites

      Abstract:First, to predict consumer sentiment polarity based on Chinese online review of catering industry, this study establishes Lasso-Logistic and Lasso-PCA models. By comparison, Lasso-PCA model is more accurate by integrating more information of variables. However, Lasso-PCA model has weaker explanatory power especially in the scenario of high dimensional data. Second, using the variable selection results of Lasso-Logistic model, we find that specialties, service attitude, and the external environment, as well as "a fly in the ointment" are the significant factors affecting the consumer's emotional polarity directly.

    • Convolution Neural Network Model Compression Method Based on Statistical Analysis

      2018, 27(8):49-55. DOI: 10.15888/j.cnki.csa.006481

      Abstract (1459) HTML (734) PDF 1.47 M (1754) Comment (0) Favorites

      Abstract:Aiming at the problem of convolutional layer parameter redundancy and low operation efficiency in convolutional neural network, a convolution neural network (CNN) model compression method based on statistical analysis is proposed in this paper. On the premise of ensuring a good ability of convolutional neural network to process information, the well-trained convolution neural network model is compressed by pruning the convolution kernels which have less influence on the whole model in the convolution layer, meanwhile, reducing the parameters of CNN without losing the model accuracy so as to reduce the amount of computation. Experiments show that the proposed method can effectively compress the convolution neural network model while maintaining a good performance.

    • Application of Mobile Edge Computing Technology in High Speed Rail Communication Network

      2018, 27(8):56-62. DOI: 10.15888/j.cnki.csa.006507

      Abstract (2235) HTML (1441) PDF 1.16 M (2153) Comment (0) Favorites

      Abstract:With the increasing popularity of high-speed rail, the establishment of mobile data channels between high-speed trains and ground to meet the needs of vehicle data transmission and passenger Internet access has become an increasingly urgent issue. The existing solutions of GSM-R and LTE-R still exist such problems as small bandwidth, large delay, and unstable transmission. To this end, this study uses Mobile Edge Computing (MEC) to optimize the high-speed communication network. The main idea is to deploy an MEC server in the high-speed rail and the base station. Under two levels of MEC server cooperation, the solution can reuse air interface links, improve wireless transmission stability, and reduce latency. Finally, users will get a better experience. The actual network test results show that the scheme can significantly improve the transmission rate and reduce the transmission delay.

    • Enterprise Knowledge Management System Based on Hadoop

      2018, 27(8):63-69. DOI: 10.15888/j.cnki.csa.006485

      Abstract (1846) HTML (1607) PDF 1015.17 K (1746) Comment (0) Favorites

      Abstract:Hadoop-based knowledge management systems can enhance corporate ability to store and process massive knowledge and realize knowledge discovery oriented to management decisions. Through studies on Hadoop platform and its characteristics, the advantages of Hadoop-based knowledge management systems were analyzed, a Hadoop-based corporate knowledge management framework model was built, and a logical framework was designed for Hadoop-based knowledge management systems. Eventually, detailed design was done on the knowledge recommendation module by applying MapReduce-based collaborative filtering algorithm. With knowledge management systems being built based on Hadoop and other big data technologies and recommendation system technologies, knowledge management can be personalized and intelligentized, and the scalability and economic demands of corporate knowledge management systems can be met as well.

    • Learning System of 3D Animation Automatic Generation

      2018, 27(8):70-74. DOI: 10.15888/j.cnki.csa.006472

      Abstract (1675) HTML (972) PDF 714.88 K (1858) Comment (0) Favorites

      Abstract:The automatic generation system of mobile phone 3D animation realizes that the 3D animation of SMS messages is generated automatically with the computer-aided technique. This study designs and implements a learning system for the automatic generation system of mobile phones 3D animation. The learning system helps the original system generate more satisfying animations for the users by building a random decision forests model which uses user evaluation as classification result. At the same time, the learning system can constantly update the learning model with the operation of the automatic generation system of mobile phones 3D animation, ultimately, to accomplish the "never-ending learning" ability.

    • Intelligent Auxiliary Correction System for X-Ray Powder Diffractometer

      2018, 27(8):75-80. DOI: 10.15888/j.cnki.csa.006495

      Abstract (1455) HTML (688) PDF 1.09 M (1272) Comment (0) Favorites

      Abstract:X-Ray powder Diffraction (XRD) is an exact instrument for material research with complex physical structure. Thus, a small hardware deviation will affect the quality of diffraction data, and the deviation correction of its hardware equipment will effectively guarantee the correctness of the data. Therefore, the deviation type recognition should be carried out before correcting the deviation. In general, XRD equipment is calibrated by the the equipment manufacturer expert using the special instruments to inspect item by item. This calibration highly depends on personnel experience, and the process is complicated and unefficient. This paper discusses the XRD error types and causes, and develops a intelligent auxiliary correction system for XRD based on classification and prediction, to improve the efficiency of error recognition. The system uses the general computer hardware, analyzes the standard sample diffraction data, then identifies the instrument status and assists correction automatically by feature extraction, model training, and other steps. The results show that the system can quickly identify the deviation and achieve the purpose of auxiliary correction.

    • Customized Network Security Service for Health Cloud

      2018, 27(8):81-86. DOI: 10.15888/j.cnki.csa.006506

      Abstract (2263) HTML (701) PDF 1.10 M (1997) Comment (0) Favorites

      Abstract:In view of the traditional cloud computing network security model can not satisfy the different network security needs of cloud users, this study proposes a customized network security service solution for health cloud. This scheme firstly fills in the security spec according to the health cloud user's network security requirements. Then, encrypt the security spec. Finally, the customized network security service system automatically analyzes and processes to establish effective security rules and ultimately protect the cloud user security in the medical health cloud platform. Compared with the traditional method, the results show that the proposed method is effective in improving performance.

    • EAST Plasma Profiles Data Visualization

      2018, 27(8):87-91. DOI: 10.15888/j.cnki.csa.006448

      Abstract (1580) HTML (578) PDF 763.19 K (1948) Comment (0) Favorites

      Abstract:Experimental Advanced Superconducting Tokamak (EAST) has preliminarily set up more complete database of analyzing data. Compared with the primary database of storing the raw acquisition signals, the processed database includes processed data with physical meanings from EAST subsystems (diagnostics, auxiliary heating, fueling, etc.). In order to make scientific researchers view and analyze the plasma profile data more conveniently, we use the Python and PyQt to develop user graphics software. The software can analyze and compare plasma profile data from different diagnosis systems, use graphs to show data and change the graphic attributes, etc. The system also realizes the visualization of analysis plasma profiles between shots to aid plasma operation during the experiment. This paper introduces the design scheme and realization method of visualization system of plasma profile data, and also gives the initial test results of the system.

    • Mobile Patrol System in Protected Areas

      2018, 27(8):92-96. DOI: 10.15888/j.cnki.csa.006455

      Abstract (2217) HTML (1371) PDF 814.59 K (2192) Comment (0) Favorites

      Abstract:At present, the patrolling work of many natural reserves still uses the traditional patrolling way, which has seriously affected the efficiency of patrolling work. In this study, according to the needs of actual patrol work in natural reserves, the mobile patrol system of protected areas is constructed based on the main technologies of Android mobile development, Java Web development, KML, and LBS to replace the traditional patrolling way. The system includes the mobile terminal acquisition system and Web-side management system. Patrol staffs use the mobile terminal acquisition system to improve the efficiency of patrol information collection; protected area managers use the Web system to conduct efficient management of patrolling data, and timely learn the work of patrolling staff and real-time location. The system has been used in a multiple protected areas and has achieved good results.

    • Intelligent Shared Parking System Based on REST Architecture

      2018, 27(8):97-102. DOI: 10.15888/j.cnki.csa.006461

      Abstract (1915) HTML (1136) PDF 964.71 K (1776) Comment (0) Favorites

      Abstract:In order to effectively alleviate the contradiction between the parking difficulties and a large number of parking spaces with low utilization, this paper presents a solution to the shared parking system from the perspective of the Internet of Things which adopts REST architecture to connect the independent devices such as users' mobile phone app, information storage server, intelligent parking lock, etc., supporting real time discovery of parking spaces, share time sharing, online booking, smart navigation, and unmanned supervision. This forms another case of sharing economy.

    • Intelligent Gas Data Management System Based on CoAP

      2018, 27(8):103-107. DOI: 10.15888/j.cnki.csa.006470

      Abstract (1852) HTML (1813) PDF 826.30 K (1649) Comment (0) Favorites

      Abstract:In recent years, the technology of the Internet of Things has developed rapidly. How to integrate the data of the intelligent Internet of Things with the enterprise information system is the core technology of the Internet of Things. In this paper, we analyze the two existing schemes of SOA architecture, the integration of enterprise information system and the underlying sensor network, and propose a hybrid scheme to avoid the shortcomings of the existing schemes. According to the actual background of Gas Co.'s remote meter reading, an intelligent gas data management system based on CoAP is realized. The results show that the design scheme is feasible. This scheme has the characteristics of strong expansibility and high transmission efficiency.

    • EAST Instant Messaging System

      2018, 27(8):108-113. DOI: 10.15888/j.cnki.csa.006482

      Abstract (1495) HTML (749) PDF 1022.83 K (1832) Comment (0) Favorites

      Abstract:As EAST discharge experiments will produce a large amount of experimental data, it becomes necessary to organize these experimental data to make users can efficiently obtain them and understand the progress of experiments. In this study, we propose the EAST instant messaging system based on the Openfire and Spark framework, which can communicate instantly, share files, and subscribe the experimental results. But, we focus on the subscription and push function. The Subscription/Push includes two parts:front-end subscription website and the back-end push system. Researchers can subscribe their experimental data in website. The read thread of the back-end push system will read data from MDSPlus once new experimental data generates, then the data processing thread will handle them and package into Messages with the users' JID, finally the send thread will send Messages to the server. The system is designed to solve the problem that researchers cannot obtain the progress and the results of experiments in time and it provides a good commu- nication platform for EAST experiment.

    • Application Layer Communication Protocol of Can Bus in Tracking and Positioning Fire Protection System

      2018, 27(8):114-118. DOI: 10.15888/j.cnki.csa.006476

      Abstract (2181) HTML (1623) PDF 923.73 K (2079) Comment (0) Favorites

      Abstract:In order to improve the reliability and stability of the automatic tracking and positioning fire protection system, the data transmission of the fire fighting system is proposed by using CAN bus. The architecture and communication of the fire control system are introduced, and the CAN bus communication circuit structure and principle are analyzed. On the basis of CAN 2.0 protocol B frame structure, the extended frame ID form and meaning of the message identifier are studied, the formulation, implementation method, and design examples of application layer communication protocol are given based on the CAN bus. The application results show that the design can not only meet the actual requirements of data transmission reliability and real-time performance, but also have extremely strong portability and expansibility.

    • E-Business Website Illegal Behaviors Supervision Platform

      2018, 27(8):119-125. DOI: 10.15888/j.cnki.csa.006512

      Abstract (1624) HTML (1579) PDF 1.63 M (2768) Comment (0) Favorites

      Abstract:According to the manual recognition and identification difficulty of the E-business illegal behavior, an E-business supervision platform is developed through the use of the internet search engine, text retrieval, data mining, and bloom filtering technologies. This platform supports the identification and establishment of the E-business economic subject database, the illegal E-business behavior and advertising behavior search. It also realizes the electronic identification display and anti-fake detection. This platform shows better efficiency and superiority than manual working in practice.

    • Performance Testing on Android Application Based on GUI

      2018, 27(8):126-131. DOI: 10.15888/j.cnki.csa.006492

      Abstract (1471) HTML (1213) PDF 845.84 K (1589) Comment (0) Favorites

      Abstract:Mobile applications have become increasingly popular over the recent years. As shown in the latest statistics, over 85.1% of mobile phones is based on Android operation system. Due to the Android's open-source character, Android application is becoming increasingly popular. Nevertheless, how to ensure program quality has become a severe problem. Previous work focused on doing static or dynamic analysis on the code to find performance problem. Yet they did not focus on the combination of performance testing and oriental testing. Therefore, in this paper we presents a GUI-based software automated performance testing framework. By parsing pages, we try to find the APP states as much as possible, and then analyze the log files to find out the APP performance problems and recover the problem scenarios. To ensure the operability of our tool, we use Java to set up this automated testing framework for Android mobile software. We download mobile applications from the open source community named F-Droid for experiments. The result shows that our technology can find more state of the APP and reveal performance problems in the running process.

    • Personalized Academic Article Recommendation with Tagged Convolutional Nets

      2018, 27(8):132-137. DOI: 10.15888/j.cnki.csa.006494

      Abstract (2241) HTML (1621) PDF 1.12 M (2049) Comment (0) Favorites

      Abstract:In user-oriented article collection system, users construct their collection sets by adding articles which they are interested in, studying why users desire specific articles and find the specific articles is particularly an interesting issue in social science. In this paper, we present the prediction users' preference on articles by considering content information, i.e. semantic information and tags. In this study, we propose a new model which jointly performs deep representation learning for the content information and collaborative filtering for the collection matrix. Extensive experiments on the real-world datasets show that it can significantly advance the state-of-the-art.

    • Application of Genetic Algorithm to Single Machine Batch Scheduling Problem with Energy Cost Consideration

      2018, 27(8):138-145. DOI: 10.15888/j.cnki.csa.006489

      Abstract (1358) HTML (1051) PDF 1.47 M (1367) Comment (0) Favorites

      Abstract:Energy cost has become a critical factor in production scheduling where trade-off between makespan and total energy consumption should be considered. In this study the genetic algorithm is applied to the single machine batch scheduling problem with energy cost consideration and a model which simultaneously optimizes the makespan and total energy cost was proposed. By using the genetic algorithm, a set of non-dominated solutions are obtained in the situation of Considering Energy Consumption (CEC) and Ignoring Energy Consumption (ICE) respectively and the algorithm's efficiency was guaranteed by optimizing the batch and improving the selection of the genetic operators. The experimental results show that the solution is obtained under the CEC has better effectiveness than that under the IEC. Moreover, the performance of CEC is getting better obviously when the problem size and job power increase.

    • Multi-Instance Multi-Label Algorithm Based on Label Correlation

      2018, 27(8):146-152. DOI: 10.15888/j.cnki.csa.006490

      Abstract (1738) HTML (708) PDF 1.04 M (1361) Comment (0) Favorites

      Abstract:Multi-Instance Multi-Label (MIML) learning is a novel machine learning framework in which an instance is described by multiple instances and associated with multiple labels. This framework has become a hot topic in the field of machine learning because of its excellent expressive ability for polysemous objects. The most direct way to solve the MIML classification problem is the degradation strategy, it takes the multiple instance learning or multiple label learning as a bridge, transforms the MIML problem into a series of binary classification problems. However, the correlation information among labels will be lost in the degradation process, which will affect the classification result. Based on these problems, this study proposes the MIMLSVM-LOC algorithm. The algorithm combines the improved MIMLSVM algorithm with a local label correlation method ML-LOC which considers the correlation information among labels in the training process. The algorithm first improves the K-medoids clustering algorithm in the MIMLSVM algorithm, and then uses the mixed Hausdorff distance to transform each instance packet into an instance, which degradate the MIML problem. Then, the ML-LOC algorithm is used to continue the classification work. In the experiment, the comparison experiment with other MIML algorithms, the result shows that the improved algorithm has better performance than other classification algorithms.

    • Semantic Relationship Recognition of Oil Documents Based on Improved Word Vector

      2018, 27(8):153-158. DOI: 10.15888/j.cnki.csa.006480

      Abstract (1310) HTML (542) PDF 1.16 M (1248) Comment (0) Favorites

      Abstract:Semantic relationship recognition is the process of document processing and is used to identify the semantic relations contained in the process, which is an important part of the construction of ontology. In the process of constructing petroleum field ontology, the semantic relationship identification is more difficult because the documents in the petroleum field have their unique characteristics. The current semantic recognition algorithm is mainly based on association rules' recognition algorithm, but there is no field-specific orientation. By analyzing the characteristics of petroleum documents, this study proposes a semantic relationship recognition algorithm for petroleum documents based on improved word vector. Based on the Continuous Bag-Of-Words (CBOW) model, this study carries out expanded model training on petroleum terminologies and introduces negative sampling and subsampling techniques to improve the training accuracy and efficiency. Feature vectors are used in training the Support Vector Mechine (SVM) classifier for semantic relationship recognition. The experimental results show that the word vectors trained by this method can accurately identify the semantic relations contained in documents in the petroleum field and have obvious advantages.

    • Sentiment Analysis of Commodity Reviews Based on LSTM

      2018, 27(8):159-163. DOI: 10.15888/j.cnki.csa.006483

      Abstract (2316) HTML (4104) PDF 932.24 K (2708) Comment (0) Favorites

      Abstract:With the development of e-commerce, a large number of reviews of goods have been produced. According to short text features of commodity reviews, emotion classification based on sentiment dictionary needs a lot of emotion database resources, and machine learning method needs complex artificial design features and feature extraction process. This study adopts the short and long term memory network (Long Short Term Memory, LSTM) text classification algorithm sentiment analysis. Firstly, the text word vector is introduced into LSTM network by using Word2Vec and word segmentation technology, and finally the classification model is obtained by Dropout algorithm. Experiments show that:in deep learning based sentiment analysis of commodity reviews, the unique characteristics of short-term memory network have sound results on commodity reviews sentiment classification, the accuracy of classification is more than 99%.

    • Website Authority Prediction Based on Deep Learning

      2018, 27(8):164-169. DOI: 10.15888/j.cnki.csa.006474

      Abstract (1782) HTML (609) PDF 1015.43 K (1454) Comment (0) Favorites

      Abstract:Website authority is generally measured by external links. The more high-quality external links are, the more authoritative the website or web page itself is. Evaluation website authoritative algorithm has PageRank and so on. However, the impact of such algorithms on the authority of the website is selective, making this method has some drawbacks. This study uses the method of deep learning, by mapping search terms and URLs into vectors, and then calculates the similarity between two vectors to judge the authority of different websites under a certain search term. The website with high similarity of calculation results is referred to as an authoritative site under the search term, so we can use another view to measure the authority of website. By comparing two different model experiments using Word2vec and LSTM, the experimental results on open datasets show that it is effective to use both models, and LSTM model is better than Word2vec model.

    • Algorithm for Circle Clipping Based on Arbitrary Polygon Window

      2018, 27(8):170-175. DOI: 10.15888/j.cnki.csa.006437

      Abstract (1677) HTML (840) PDF 1.27 M (1388) Comment (0) Favorites

      Abstract:For the problem of the circle clipping against arbitrary polygon window, the more comprehensive and effective clipping algorithm is proposed in this study. First, according to x-scan line algorithm, the spatial relationship between the circle and the polygon window is determined. Next, for the case of the polygon window and the circle intersection, the intersect points of the circle and each side of the polygon window are calculated in the counterclockwise direction and sorted correctly. At last, according to the relationship between two points, determining to draw a line or a circle arc. The whole circle clipping is obtained. The result expresses that the algorithm can be comprehensive and effective to complete circle clipping.

    • Image Recognition Technology for Transmission Line External Damage Based on Depth Learning

      2018, 27(8):176-179. DOI: 10.15888/j.cnki.csa.006458

      Abstract (2129) HTML (1359) PDF 731.11 K (1903) Comment (0) Favorites

      Abstract:In the power system, it is very important to identify and eliminate the hidden dangers of transmission lines to ensure the power system's security. Image recognition technology is an effective method to identify the risk of breaking out. According to the hidden breaking danger recognition problem, this study proposes a depth model by training the convolutional neural network algorithm. According to the anti breaking characteristics of risk image on the existing depth network structure are improved by increasing the ROI pool layer and modifying the loss function. A large number of training samples are used to get the robust model test when the measured image is first in generated candidate region, then the detection and identification for each candidate region are carried out, to detect potential risks to break out in a complex background. The experimental results show that this method can effectively identify the hidden danger of transmission lines.

    • Scheduling Problem with Sigmoid Satisfaction Emergency Vehicle Based on Whale Swarm Algorithm

      2018, 27(8):180-186. DOI: 10.15888/j.cnki.csa.006467

      Abstract (1667) HTML (1096) PDF 1.31 M (1530) Comment (0) Favorites

      Abstract:Considering the different levels of large-scale urgent disaster, Sigmoid time satisfaction function is proposed to evaluate the rescue effect. It builds the shortest rescue path and maximum satisfaction of average time vehicle scheduling model. Chaotic Whale Swarm Algorithm is designed based on Chaos sequence search operator. the improved algorithm is applied to optimize three groups of experimental cases of different sizes, and the results are compared with those of the simulated annealing algorithm and original Whale Swarm Algorithm. Results show that there is little difference in performance when the handling smaller scale of vehicle scheduling, while the improved algorithm gain more advantage over the simulated annealing algorithm and original algorithm as the handling scale getting larger. Chaotic Whale Swarm Algorithm is effective method to deal with emergency vehicle scheduling problem.

    • Research on Device Defect Recommendation Based on Tag Technology and Entropy Weight Method

      2018, 27(8):187-192. DOI: 10.15888/j.cnki.csa.006459

      Abstract (1567) HTML (701) PDF 889.11 K (1476) Comment (0) Favorites

      Abstract:According to the equipment defect interconnection between the Production Management System (PMS) and the Operation Management System (OMS) requires the PMS maintenance staff to choose the defects personally, resulting in staff workload increased significantly and the extent of unreasonable data interaction. At the same time, the incomplete degree of interconnection is gradually increased. The study proposes a recommendation defects method based on tag technology and entropy coefficient method. Firstly, the forward maximum matching algorithm and edit distance and rule database technologies are being used for tagging defects identification, and then using entropy weight method to evaluate the label, in order to achieve the intelligent recommendation to relevant personnel. The experimental results show that the implementation of the proposed method greatly reduces the workload of the relevant personnel, and improves the accuracy of the defect information recommendation.

    • Recognition Algorithm of Moving Target in Complicated Scenes

      2018, 27(8):193-197. DOI: 10.15888/j.cnki.csa.006465

      Abstract (1373) HTML (1538) PDF 809.53 K (1595) Comment (0) Favorites

      Abstract:Target recognition is the basic purpose of computer vision, it is also one of the key components in the field of artificial intelligence.With the advent of the information age, the popularity of video capture tools, massive video data to human identification has brought great challenges. At this stage, video recognition technology has been widely used in simple scenes such as intelligent transportation field and production quality inspection field. How to realize the target recognition and detection from complex scenes has become a more important and difficult issue. In response to this problem, this paper presents a moving target recognition algorithm in complex scenes. First, an improved optical flow algorithm is proposed to mark the moving target region quickly by time series and spatial pixel changes; Secondly, the sliding window of the target area is detected to match the model of each part of the human body, and the feedback information is modeled by a tree structure. Through experiments, this method can detect faster than detection algorithm based on depth learning while ensuring high accuracy, and can meet the requirements of real-time monitoring.

    • Detection of Random Generated Names Using Recurrent Neural Network with Gated Recurrent Unit

      2018, 27(8):198-202. DOI: 10.15888/j.cnki.csa.006466

      Abstract (1403) HTML (693) PDF 879.77 K (1541) Comment (0) Favorites

      Abstract:Random domain names refer to the names generated by domain generation algorithms, which are widely used by the malware of the computer network system. The detection of random domain names is the basic work of the traffic filtering operation of the domain name system. The traditional method of detecting random domain name is not ideal and the precision and recall are low which will lead to erroneous judgement in attack traffic filtering. In this paper, the random names detection model are built based on recurrent neural network with gated recurrent unit. In this model, domain names are converted to vectors at first, then GRU are adopted to learn features automatically and which will be taken by the neural network to compute the class scores. Compared to traditional methods, this method is able to extract features without human help and which will reduce the time cost of feature extraction. This method performs better in the experiments of the algorithm generated data and the real world data than traditional models.

    • Dynamic Threat Data Model and Visual Perception of Power Grid Fused Ontology Theory

      2018, 27(8):203-208. DOI: 10.15888/j.cnki.csa.006473

      Abstract (1358) HTML (587) PDF 1.17 M (1474) Comment (0) Favorites

      Abstract:Visualization of network security threats, deep integration of network status, and attack patterns, combine the security situational awareness and visualization technology, can realize the visualization representation of the global network under the trusted state. The technologies of network threat visualization still have the problem of limited representation ability of traditional data models, and the low availability of state feature redundancy and dispersion. In this study, dynamic threat data model and visual perception are proposed, which combines the ontology theory and the situation evolution. Unified attack behavior model with three solid states can help improve the problem of unclear description of network security characteristics. By designing deep content detection based on ontology features, the tight relationship characteristics set reduces the redundancy of security features. The refined network threat data will pass through the situation ladder to achieve the smooth gradient of the graphical representation of the attack behavior. The tests on Zunyi Power Supply Bureau of Guizhou Province verify the proposed method to improve the network security threat monitoring error by 4%.

    • Social Image Caption with Visual Attention and User Attention

      2018, 27(8):209-213. DOI: 10.15888/j.cnki.csa.006501

      Abstract (1329) HTML (893) PDF 799.97 K (1565) Comment (0) Favorites

      Abstract:Image captioning has attracted much attention in the field of machine learning and computer vision. It is not only an important practical application, but also a challenge for image understanding in the field of computer vision. Nevertheless, existing methods are simply rely on several different visual features and model architectures, the correlation between visual features and user tags has not been fully explored. This study proposes a multifaced attention model based on user tags and visual features. This model can automatically choose more significant image features or contain the user semantic information. The experiments are conducted on MSCOCO dataset, and the results show that the proposed algorithm outperforms the previous methods.

    • Design of Distributed Digital Signal Processing Algorithm Library Based on Spark

      2018, 27(8):214-218. DOI: 10.15888/j.cnki.csa.006508

      Abstract (2234) HTML (1143) PDF 783.59 K (1675) Comment (0) Favorites

      Abstract:The traditional DSP and FPGA-based digital signal processing technology is more suitable for real-time signal processing. It is limited by the size and frequency resolution of the data. So it unsuitable for applications of off-line data processing, data analysis and data mining under large-scale data. Currently the industrial big data analytics platforms can use Spark as a computational engine for real-time signal processing and off-line signal processing acceleration. However, there is a lack of mathematical solutions such as digital signal processing for distributed parallel computing engines. Consequently, this paper presents a library of distributed digital signal processing algorithms based on Spark, which provides a support for the analysis of industrial big data application scenarios. This paper describes the architecture of the algorithm library design and takes the FFT algorithm and DFT algorithm as examples to introduce the distributed implementation of the traditional digital signal algorithm in Spark. Finally, this paper persents a test and analysis for this algorithm library. The results show that the algorithm library can correctly accomplish the function of digital signal processing and it can fulfill the industrial big data analysis platform for large-scale data sets for digital signal processing needs.

    • Solving Traveling Salesman Problem Based on Improved Firefly Algorithm

      2018, 27(8):219-225. DOI: 10.15888/j.cnki.csa.006497

      Abstract (1832) HTML (700) PDF 1.32 M (1824) Comment (0) Favorites

      Abstract:Traveling Salesman Problem (TSP) is an oldest combinatorial optimization problem, and Firefly Algorithm (FA) shows excellent performance to complicated function optimization. Hence, in this study, we used improved FA to solve TSP. Firstly, after the characteristics of TSP are analyzed, and the method of integer encoding is adopted to set the position of fireflies. Then, the logarithmic adjustment factor is introduced in the standard FA. Meanwhile, we combine the crossover, mutation, and reverse operation in Genetic Algorithm (GA) to improve the population diversity and search ability of each iteration, and it is applied to solve TSP. Finally, the numerical experiments show that the proposed algorithm has faster convergence speed and optimization effect.

    • Sentiment Analysis of Short Messages Based on Attention Mechanism and Convolution Neural Network for Mobile Animation

      2018, 27(8):226-231. DOI: 10.15888/j.cnki.csa.006468

      Abstract (1458) HTML (615) PDF 845.26 K (1559) Comment (0) Favorites

      Abstract:In 2008, ZHANG Song-Mao, a researcher of Chinese Academy of Sciences, proposed the application of 3D animation automatic generation technology for mobile phone SMS. The sentiment analysis of SMS is an important part of the 3D animation automatic generation system. At present, the method used in the system is a traditional machine learning method, which has a low accuracy and cannot achieve a practical purpose. In recent years, deep learning has achieved good results in the task of sentiment analysis. Convolutional neural network can automatically extract the semantic and sentiment features of text messages, and attention mechanism can automatically obtain weighting information for words. Therefore, this study proposes to apply the attention mechanism and convolutional neural network in deep learning to the classification of sentiment analysis in the system of SMS automatic generation. Experiments show that the convolutional neural network based on attention mechanisms has significantly improved the accuracy, recall rate, and F-value than the previous methods.

    • Performance Prediction Model for Spark Based on Key Stages Analysis

      2018, 27(8):232-236. DOI: 10.15888/j.cnki.csa.006469

      Abstract (1832) HTML (1029) PDF 888.59 K (1486) Comment (0) Favorites

      Abstract:Spark is widely used as a computing platform for large data processing, reasonable allocation of cluster resources plays an important role in the operation of Spark performance optimization. The performance prediction is the basis and key of cluster resource allocation optimization, thus we put forward a Spark performance prediction model in this paper. This paper selects the job execution time as a measure indicator of Spark performance, and put forward the concept of key Stage of Spark job. Finally, we built the model by analyzing relationships between the key Stages and the amount of input data through running a small quantity of data. The experimental results show that the model is effective

    • Ultra-Wideband Location Algorithm Based on LS_SVM

      2018, 27(8):237-240. DOI: 10.15888/j.cnki.csa.006475

      Abstract (1560) HTML (749) PDF 898.44 K (1401) Comment (0) Favorites

      Abstract:Designated to solve the problem of large positioning error caused by large ranging errors in the non-line-of-sight (UWB) environment, a tracking error based on Least Squares Support Vector Machine (LS_SVM) is proposed. The method divides the indoor area into several equal small areas, establishes the non-linear relationship between the eigenvalues of the sampled signals and the node locations in each area, classifies and regulates them by using LS_SVM. For non-line-of-sight distance measurement results, a smaller weight is given. Experimental results show that the error of K-Nearest Neighbors (K-NN) is improved by 10% within 7 cm, which shows that this algorithm can effectively improve the positioning accuracy.

    • Engineering Practice of Smart Mini Car Based on ARM and Embedded Linux

      2018, 27(8):241-246. DOI: 10.15888/j.cnki.csa.006510

      Abstract (2133) HTML (3665) PDF 1.12 M (2214) Comment (0) Favorites

      Abstract:The engineering practice of mini smart car with WiFi wireless remote control function based on ARM and embedded Linux is discussed. The process, method, and experience of engineering practice are introduced. The ARM11 chip is selected as the central control CPU. The WiFi wireless communication module on smart car is used to communicate with remote Android smart phone to receive controlling commands. Controlling commands are sent by the smart phone through the novel ‘gravity sensing’ or traditional ‘key pressing’ mode. Once the commands are received by the ARM chip, it will control the motors via GPIO and drive four wheels to make according actions (moving forward/backward, turning and rotating), and force the car to move in a straight line via PID algorithm. The video data can be collected by camera and the real-time human faces detection is implemented by computer vision algorithms.

    • Improved Time-Delay Estimation Algorithm for Mutual Power Spectrum

      2018, 27(8):247-253. DOI: 10.15888/j.cnki.csa.006486

      Abstract (1695) HTML (1916) PDF 1.26 M (2094) Comment (0) Favorites

      Abstract:The sound source location of the microphone array has been a hot topic in the field of array signal processing. The time-delay estimation method, which is represented by mutual power spectrum phase estimation (CSP), is widely used because of its simple principle, small computation, and easy implementation. Although the CSP algorithm has a good estimation effect in the high SNR environment, the accuracy is drastically reduced when the SNR is low and the acoustic scene is much complicated. In order to solve this problem, this study improves the CSP algorithm. By filtering the time-delay estimation results of the CSP algorithm, the unreasonable delay value is eliminated, the algorithm parameters are updated and estimated again to obtain reasonable delay value, and through multiple frame signal to obtain the time-delay and position information of the sound source. In order to verify the effectiveness of the proposed algorithm, this paper experimentally validates in Matlab and real environment respectively, and the results show that the enhanced CSP algorithm has improved the accuracy of the time-delay estimation compared with the original algorithm.

    • Booking Service APP for Elevator Based on Android

      2018, 27(8):254-258. DOI: 10.15888/j.cnki.csa.006430

      Abstract (2564) HTML (1006) PDF 890.08 K (1909) Comment (0) Favorites

      Abstract:This study designs a mobile APP of elevator booking system, designating to solve the problems of the reservation site and booking method immutable. The system accesses the key technology to realize the remote client and field device data sharing and remote monitoring by using the elevator data acquisition, control terminal, and server database. The field devices establish a link with the server link using RS485 bus, through the embedded elevator controller. The APP completes the client real-time updates data according to the elevator running characteristics of the development of mobile applications. The design is based on the Android system, with equip went of the elevator location display, the number of elevator monitoring, remote call, and elevator to remind the station and other functions. These functions can choose the suitable time for users according the elevator using status.

    • Automatic Generation Method of Prose Poem Based on Recurrent Neural Network

      2018, 27(8):259-264. DOI: 10.15888/j.cnki.csa.006318

      Abstract (2458) HTML (810) PDF 1.01 M (1705) Comment (0) Favorites

      Abstract:Aiming at the automatic generation of poem, a temporal text generation method using recurrent neural network is proposed. After building a word set according to the existing corpus, a number of keywords is given, and the first sentence is generated by expanding the keywords based on the word set constructed by clustering model. On the basis of the first sentence determined, the generated content is compressed by the context model and the feature is extracted, and finally the content of the previous context is passed to the generation model to realize the subsequent sentence generation. In order to achieve the mapping between the upper and lower sentences, the first sentence of the process is a vocabulary expansion, and the context model can be a good grasp of the context. BLEU automatic evaluation method and manual evaluation method are used to establish a more standard evaluation system. The results approve the effectiveness of the method.

    • Research on Privacy Analysis of Mobile Implicit User Behavior Data

      2018, 27(8):265-269. DOI: 10.15888/j.cnki.csa.006424

      Abstract (1490) HTML (745) PDF 700.85 K (1255) Comment (0) Favorites

      Abstract:Smart devices bring convenience to people and also record a large number of user habits, location, access content, and other private information. In this study, considering the way of user behavior data collection, data processing and storage, we design a data acquisition system of user behavior data, and collect a lot of non-explicit identity data on the smart terminal equipment, including network traffic information and screen status information. Through analyzing and processing the data, we found that this non-explicit user identity information can effectively identify the user identity, and can figure out social relationships between the users. The experiment show that, based on the explicit non-identifying information to protect user privacy data has important significance and great value.

    • Research on Noise Impact on Quality of Pet Images

      2018, 27(8):270-275. DOI: 10.15888/j.cnki.csa.006511

      Abstract (2655) HTML (902) PDF 1.19 M (2061) Comment (0) Favorites

      Abstract:The image noise affected the results of quality estimation in PET image was investigated in this research. The best image quality can be used to find the adequate iteration numbers in PET image reconstruction process. Two phantoms, Huffman and Utah, were used in this research. The simulation was processed by using a Monte Carlo algorithm to simulate phantom in a Siemens ECAT PET scanner respectively. Flowering that an MLEM iterative algorithm named OSEM was used for reconstruction. The noise level on image was estimated by standard deviation calculation. The image quality evaluation used the SSIM and the PSNR two image quality indices. The noise in PET images is ascending with the iteration numbers increasing but the image quality contradicts to the numbers of iteration for a Huffman phantom. The image quality of filtered images, by using average filter, is significantly lower than the original images. This effect indicated that the noise affects the image quality calculations. Optimal quality of image appears later than the unfiltered image, which shows that the rising noise results in the optimal image detection. A Utah phantom was also used to examine this effect and got equal results. The noise in image estimated accurately the truly optimal iterative image can be finding easily.

    • Formal Research of UML Sequence Diagram Based on Temporal Description Logics

      2018, 27(8):276-280. DOI: 10.15888/j.cnki.csa.006493

      Abstract (1572) HTML (648) PDF 750.97 K (1423) Comment (0) Favorites

      Abstract:In Unified Modeling Language (UML) superstructure standard, the semantics of sequence diagrams are formal defined by natural language, it is a semi-formal language and it can not make formal analysis and proof to system's interbehavior. Aiming at the problem that UML sequence diagram is not able to formal description, according to the temporal characteristic of UML sequence diagram, this study puts forward an six-tuple formalization method based on adding the interaction operators in UML sequence diagram. Temporal Description Logics (TDLs) are proposed by the temporal extending the Description Logics (DLs), which are the formal specification of the dynamic and temporal semantics. The semantic of TDLs is proposed by temporal operators of TDLs. Taking UML sequence diagram of the C Language executing process as an example, a formal methodology description is given out. Examples verify the feasibility of this method.

    • Recognition of Aluminum Casting Based on Texture Feature and SVM Classifier

      2018, 27(8):281-285. DOI: 10.15888/j.cnki.csa.006527

      Abstract (1634) HTML (1304) PDF 747.55 K (1315) Comment (0) Favorites

      Abstract:With the growth of the global economy and the widespread use of aluminum profiles, the global consumption of aluminum castings has been increasing year by year. Due to the different applications, there are a variety of aluminum castings, they have different shapes, structures, colors, textures and so on. As an important aspect of the image processing application, this study analyzes the features of aluminum castings, extracts the texture features of the image by using the gray level co-occurrence matrix and Gabor wavelet transform, respectively, and compares them with the SVM classification algorithm of SVM feature classification, test recognition accuracy, experimental results were compared for the classification of aluminum castings obtained Gabor wavelet transform using both the recognition accuracy or recognition of time on the results are the best.

    • Service Function Chain Design Based on Software Defined Security

      2018, 27(8):286-291. DOI: 10.15888/j.cnki.csa.006500

      Abstract (1697) HTML (2978) PDF 1.11 M (1564) Comment (0) Favorites

      Abstract:With the rapid development of cloud computing and its widespread application, security protection in the cloud environment has become a pressing issue. Aiming at the problem of how to flexibly and dynamically schedule virtualized security devices, this study proposes a service function chain based on software-defined security, builds the security service function chain according to user requirements, and schedules virtual security devices in secure resource pools. The experiment proves the validity of the method in this paper.

Current Issue


Volume , No.

Table of Contents

Archive

Volume

Issue

联系方式
  • 《计算机系统应用》
  • 1992年创刊
  • 主办单位:中国科学院软件研究所
  • 邮编:100190
  • 电话:010-62661041
  • 电子邮箱:csa (a) iscas.ac.cn
  • 网址:http://www.c-s-a.org.cn
  • 刊号:ISSN 1003-3254
  • CN 11-2854/TP
  • 国内定价:50元
You are the firstVisitors
Copyright: Institute of Software, Chinese Academy of Sciences Beijing ICP No. 05046678-3
Address:4# South Fourth Street, Zhongguancun,Haidian, Beijing,Postal Code:100190
Phone:010-62661041 Fax: Email:csa (a) iscas.ac.cn
Technical Support:Beijing Qinyun Technology Development Co., Ltd.

Beijing Public Network Security No. 11040202500063