Computer and Modernization

An Improved K-medoids Algorithm Based on Optimal Initial Cluster Center

DUAN Gui-qin1， ZOU Chen-song2， LIU Feng2

2019, 0(04): 1. doi:10.3969/j.issn.1006-2475.2019.04.001

Asbtract ( 212 )

PDF (800KB) ( 120 )

References | Related Articles | Metrics

Aiming at the initial clustering center of k-medoids may be too near, under-represented, or poor stability, an improved k-medoids algorithm is proposed. The ratio of sample sets average distance and samples average distance is treated as the density of sample parameters, the number of candidate representative points in the high density point set is simplified, the product of maximum distance method is adopted to choose K samples with high density and long distance as the initial clustering center, both of the representative and dispersion of the clustering center are considered also. Experimental results on the UCI data set show that compared with the traditional K-medoids algorithm and the other two improved clustering algorithms, the new algorithm not only has more accurate clustering results, but also has faster convergence speed and higher stability.

An Improved Mean Square Difference Collaborative Filtering Algorithm

RAO Yu, CHEN Guang, QIU Tian

2019, 0(04): 6. doi:10.3969/j.issn.1006-2475.2019.04.002

Asbtract ( 193 )

PDF (576KB) ( 112 )

References | Related Articles | Metrics

Traditional collaborative filtering algorithm based on the mean square difference (MSD) only considers the mean square difference value between the rating vectors when calculating the similarity, resulting in an unsatisfactory recommendation performance. To solve this problem, we propose an improved mean square difference collaborative filtering algorithm (IMSD), which integrates the cosine value and the mean square difference value between the rating vectors. Experiments on two Movielens datasets show that the IMSD algorithm improves the recommendation accuracy compared with the MSD algorithm. More importantly, we find that its generalized application is also effective. By applying the IMSD into improving two other algorithms, JAC_MSD and AC_MSD algorithms, we propose two corresponding JAC_IMSD and AC_IMSD algorithms, and find that the recommendation accuracy of both algorithms can be improved. Among all the investigated algorithms, the recommendation accuracy of the AC_IMSD algorithm is best.

Improved Random Balanced Sampling Bagging Algorithm for Network Loan Research

GUO Bing-nan， WU Guang-chao

2019, 0(04): 11. doi:10.3969/j.issn.1006-2475.2019.04.003

Asbtract ( 235 )

PDF (869KB) ( 91 )

References | Related Articles | Metrics

The data of network loan users in Internet finance has the characteristics of class imbalance, which seriously affects the performance of traditional classifiers. The random balanced sampling algorithm considers all samples equally in the process of resampling the original data set. In this paper, the performance of the sample points is fully considered in the process of balanced sampling, and it is divided into three types of samples: safe, boundary, and noisy. The corresponding sampling method is adopted for different types of samples to obtain a balanced new data set, and then the Bagging integration of the data set is performed to improve the generalization performance of the algorithm. The results show that the Improved Random Balanced Sampling(IRBS) Bagging algorithm in this paper can better classify loan users.

A Deep Learning Based Elastic Retractable Algorithm for Cloud Platform

CAO Yu, YANG Jun

2019, 0(04): 17. doi:10.3969/j.issn.1006-2475.2019.04.004

Asbtract ( 264 )

PDF (1896KB) ( 146 )

References | Related Articles | Metrics

To meet the performance demands and reduce resource consumption, researchers have proposed many elastic retractable algorithms and schemes. However, most of them only considered current states of the servers or applications that limit the applicability of algorithm and the performance of resource adjustment. This paper presents an application-oriented elastic retractable algorithm based on long-short term memory network and back-propagation neural network. The algorithm includes a workload prediction model, a response time prediction model and a resource adjusting model. With these models, the algorithm can predict the workload and response time of cloud computing applications and provide appropriate resource scheduling strategies. In order to improve the accuracy of workload prediction, we combine convolution operation with long-short term network for better data feature extraction and prediction. To improve the convergence speed of the model and effectively avoid the problem over-fitting, we use batch normalization in BP neural network. In the validation experiment, the mean absolute percentage error of the algorithm’s workload prediction is reduced to 3.4×10-4, and models of response time prediction and resource adjusting also perform well. In the actual operating environment, the algorithm can also provide appropriate adjustment suggestion to Docker container cloud platform. The results of experiments show that our algorithm had better performance in workload prediction and server adjusting than other models.

Adaptive Estimation Algorithm Based on Temporal-spatial Correlation Prediction

WANG Yan1， ZHU Juan1,2， WANG Lian-ming1， HUANG Ji-peng1

2019, 0(04): 25. doi:10.3969/j.issn.1006-2475.2019.04.005

Asbtract ( 166 )

PDF (1598KB) ( 99 )

References | Related Articles | Metrics

In order to eliminate spatial and temporal redundancy in video better and obtain motion vectors with sufficient accuracy quickly and efficiently, this paper proposes an improved adaptive rood pattern search algorithm. The algorithm uses temporal spatial domain correlation to predict the motion vector of the current block. For the edge image of the video, a fixed small step is used for cross search. For the non-edge part of the image, the search is performed from coarse to fine. The adaptive arm length is the maximum value of the horizontal and vertical coordinates of the predicted target motion vector. Compared with the traditional adaptive rood pattern search algorithm and several other classical motion estimation algorithms, the proposed algorithm enhances the accuracy of search prediction, reduces the average number of searches per block, and improves the search rate.

An Elasticity and Deadline-aware Job Scheduling Algorithm

HUANG Chun-qiu， CHEN Zhi， RONG Chui-tian

2019, 0(04): 30. doi:10.3969/j.issn.1006-2475.2019.04.006

Asbtract ( 327 )

PDF (974KB) ( 85 )

References | Related Articles | Metrics

We make an in-depth study of big data analysis jobs using the MapReduce model, and analyze the defects of the existing job scheduling algorithms. Most of the existing algorithms do not take into account these problems: the impact of resource allocation on job deadline and deadline sensitivity for different types of jobs. It is elastic jobs because the completion time of it that varies with the allocation of resource, deadline sensitivity means that different types of jobs have various degrees of strictness to deadlines. To solve above problems, we propose a flexible job scheduling algorithm based on deadline-aware(DA). The algorithm classifies jobs according to the sensitivity of deadline, based on the prediction of the overall execution time of jobs, by regulating different resource allocation strategies to change the completion time, combines with the users’ demand of deadline and the benefits of job pre-execution to planning the resource allocation and scheduling order of jobs in advance, for the sake of maximizing the overall benefits. We implement DA according to simulation experiment, and evaluate it on a 210 machine cluster using production workloads. The experiment shows that the algorithm satisfies the deadline and the overall yield increases by 2.37 times in average.

Transmission Channel Selection Method Based on Medium & Long-term #br# Transaction Electricity Quantity Checking

WANG Gang, WANG Zhi-cheng

2019, 0(04): 38. doi:10.3969/j.issn.1006-2475.2019.04.007

Asbtract ( 169 )

PDF (520KB) ( 102 )

References | Related Articles | Metrics

With the development of power grid and the continuous improvement of power demand, more and more medium and long-term transaction electricity quantity checking of state grid is conducted. Based on the analysis of the shortage of current short term checking and the demand of medium and long-term checking, this paper proposes the medium and long-term transaction electricity quantity checking model and the transmission channel selection strategy. Finally, by using a practical example, the reasonableness of the transmission channel selection strategy is verified.

Research on Using High-level Architecture to Develop Fire Control #br# Information Fusion Simulation System

WU Kai1， XU Li2， ZHU Jing1，LIU Xiao-yang1，LU Na1

2019, 0(04): 42. doi:10.3969/j.issn.1006-2475.2019.04.008

Asbtract ( 137 )

PDF (1189KB) ( 108 )

References | Related Articles | Metrics

The article proposes a new method of designing and realizing the simulation system which can be applied to the testing and evaluation of multi-sensor information fusion. In the article, we introduce the system function, design the common framework, analyze the key technologies of the system and build up a better one. By using component technology and plug-in technology, combining with easy-to-operate model generation tools, it provides an ideal platform for the research of information fusion algorithm.

Gait Optimization of Humanoid Robot Based on Deep Q Network

YUAN Wen， LIU Hui-yi

2019, 0(04): 47. doi:10.3969/j.issn.1006-2475.2019.04.009

Asbtract ( 222 )

PDF (2070KB) ( 159 )

References | Related Articles | Metrics

In order to realize the fast and stable walking of humanoid robot, and under the condition that the effective parameter combination is satisfied, a walking parameter training algorithm based on deep reinforcement learning is proposed to optimize the gait of humanoid robot. First of all, we capture the robot gait model parameters from the environment as the input of DQN. And then, DQN is used to fit the robot state-action value function. At last, by action selection strategy, we choose the gait of a robot to perform current action, at the same time produce reward function to achieve the aim of updating DQN. By selecting NAO robot as the experimental object and conducting experiments on the RoboCup3D simulation platform, the results show that using this algorithm, NAO robot can achieve stable bipedal walking.

Research on Personalized 3D Virtual Try-on System

ZHU Hong-qiang1， CHENG KAI1， CHEN Zhi1， LI Ling-jie1， TONG Jing1,2， JIANG Chao-qun1

2019, 0(04): 52. doi:10.3969/j.issn.1006-2475.2019.04.010

Asbtract ( 228 )

PDF (4798KB) ( 112 )

References | Related Articles | Metrics

In order to satisfy the user’s personalized demands and improve the virtual try-on experience, a personalized virtual try-on system is proposed, which includes auto-skinning, 3D face reconstruction based on a single image, parametric deformation of human body and penetration processing of garment model. Firstly, auto-skinning algorithm is designed with the aim to reduce the cost of making garment models for the existing virtual try-on systems. Secondly, individual needs of the customers can be met with a single image based automatic 3D face reconstruction and parametric deformation of human body. What’s more, on the basis of automatic calculation of transparent textures, cloth penetration is solved to improve fitting effects of the system. According to the experiments, the virtual try-on system introduced in this paper is able to achieve better fitting effects with lower construction and operation costs, in which case good personalized experience of virtual try-on is provided.

Research on Infrared Insulator Detection Based on Improved Fast-CNN Mode

JI Chao, HUANG Xin-bo, CAO Wen, ZHU Yong-can, ZHANG Ye

2019, 0(04): 59. doi:10.3969/j.issn.1006-2475.2019.04.011

Asbtract ( 162 )

PDF (4463KB) ( 95 )

References | Related Articles | Metrics

The detection effect of infrared image insulator strings is affected by the environment in the power grid inspection. The combination of saliency detection and improved convolution neural network (Fast-CNN network) is proposed for insulator feature detection. Firstly, superpixels are used to describe the overall information of each region, saliency features are calculated based on the characteristic covariance information of each superpixel. Then the salient features are extracted by regional modular extraction and local complexity contrast. At the same time, the salient features extracted from the two methods are respectively input into the improved Fast-CNN network for salient region detection, a dynamic adaptive pool model is proposed, the cosine window is introduced to deal with the middle layer. Finally, the characteristics of insulators are obtained through iterative training. It can avoid full graph search for the CNN model. The proposed algorithm is tested in the infrared image library， the F-Measure and the average error MAE of the proposed algorithm are better than the current popular algorithms.

Research on Affection of Association Rules to Pedestrian #br# Attributes Recognition in Surveillance Video

LI Xue1,2, GUO Hui-ming3

2019, 0(04): 65. doi:10.3969/j.issn.1006-2475.2019.04.012

Asbtract ( 171 )

PDF (1226KB) ( 119 )

References | Related Articles | Metrics

Aiming at the problem of pedestrian multi-attributes recognition under surveillance video, this paper proposes a multi-classification method combining neural network and association rules. Firstly, the attributes confidence of pedestrian in surveillance video can be obtained through Faster-RCNN detection algorithm and improved AlexNet multi-classification network. Then, it adopts Apriori association rules to deal with the training data. After combining neural network classification confidence and the results of association rules, it proposes an algorithm to optimize classification confidence. Finally, by analyzing the accuracy rate of some pedestrian attributes optimized by association rules, the results show that the effective combination of neural network and association rules can improve the accuracy of some attributes recognition.

Chinese Character Recognition in Complex Images Based on #br# Improved SURF Descriptor Features and Fuzzy Reasoning

TAO Xiao-jiao, LU Jin

2019, 0(04): 72. doi:10.3969/j.issn.1006-2475.2019.04.013

Asbtract ( 176 )

PDF (832KB) ( 82 )

References | Related Articles | Metrics

In considering the robustness of Chinese characters matching to the variations of orientation, position and brightness, this paper describes a new method based on improved Speeded Up Robust Features named SSURF to extract characters features. Firstly, the SSURF descriptors of the same class samples are matched. Then the matching rate of key points whose matching times exceed 1/2 is calculated. Finally, the mean value of SSURF descriptors of training samples and the maximum Euclidean distance between SSURF descriptors and mean values are used to establish class database.Experimental results demonstrate that the proposed method yields a better performance.

Research on BP-ANN Models of Lightning Prediction with Spatio-temporal Characteristics

LI Fen1， XIAO Jian2, LIN Zhi-qiang2, LI Zhi-peng1

2019, 0(04): 76. doi:10.3969/j.issn.1006-2475.2019.04.014

Asbtract ( 148 )

PDF (757KB) ( 105 )

References | Related Articles | Metrics

In order to improve the accuracy and learning performance of the lightning prediction model, a BP-ANN binomial classifier of lightning prediction based on incremental learning and spatio-temporal characteristics is proposed. It makes a study of historical data by incremental approach and according to spatio-temporal characteristics of data, builds many BP-ANN models, classifies the new data respectively, and then uses the majority voting to determine the category of the new data. This paper constructs three kinds of lightning prediction models, BP-ANN model based on incremental learning, BP-ANN model based on spatio-temporal characteristics, and BP-ANN model combined both. The accuracy and learning performance are tested on real lightning data set, the results show the advantages and disadvantages of incremental learning, spatio-temporal characteristics and combination of both.

Data Interaction Between Enterprise Level Business Systems Integrating #br# High Dimensional Random Matrix Data Analysis Model

YAN Bin-yuan， WEI Li-peng， ZHOU Lin-yan

2019, 0(04): 82. doi:10.3969/j.issn.1006-2475.2019.04.015

Asbtract ( 155 )

PDF (2148KB) ( 82 )

References | Related Articles | Metrics

Based on the analysis of the development status of enterprise-level business systems and the lack of normalized and standardized data sharing mechanism in system construction, an inter-system data interaction method for enterprise-level business system integrating large data analysis model and high-dimensional random matrix (LDAMMHDRM) is proposed. This method uses high-dimensional random matrix to demodulate hidden information to obtain hidden frequency sequence and constructs a large data analysis model to reduce the amount of running memory needed for data interaction between enterprise-level business systems. The simulation results show that this method can effectively improve the data interactive application ability between enterprise-level business systems. The fault tolerance of the model is very good.

ARIMA Based Fishing Economy Prediction Model and Its Optimization

CAI Ge-jing1, FU Hai-bin1, JIANG Ren-bin1, HUANG Bin2, ZHANG Zheng2, ZHANG Heng1

2019, 0(04): 87. doi:10.3969/j.issn.1006-2475.2019.04.016

Asbtract ( 163 )

PDF (686KB) ( 118 )

References | Related Articles | Metrics

As one of the important foundation of national economy, fishery is necessary to forecast. In this paper, the time series ARIMA model is used to predict the total output value of fishery, and the error analysis is made based on the model prediction results. Considering the effect of inflation on the prediction model, the model is optimized by using CPI index. Taking the total fishery output value of Jiangsu province as an example, the data from 1995 to 2014 were taken as training samples, and the model was established and optimized with CPI index. The data from 2015 to 2018 were taken as test samples, the results show that the optimization model had good prediction effect.

An Empirical Analysis of Public Participation in Social Security #br# Governance Based on Logistic Model

NIU Ju-ling， YANG Li-min， HOU Yun-xia

2019, 0(04): 92. doi:10.3969/j.issn.1006-2475.2019.04.017

Asbtract ( 148 )

PDF (888KB) ( 80 )

References | Related Articles | Metrics

Social security is closely related to the interests of the public, therefore, the governance of social security can not be separated from public participation. In order to analyze the influencing factors of public participation intention in social security governance, this paper takes the theory of knowledge, belief, attitudes and practice, and the theory of planned behavior as the theoretical basis, and designs relevant issues for field investigation. Using Logistic regression method, it is found that the individual characteristics of the public have different effects on the public’s willingness to participate in social security governance. The public’s cognition, attitude, subjective norm and perceptual behavior control have positive influence on the public’s willingness to participate in social security governance. So we can explore the dimensions and ways of public participation in social security governance, and provide reference for public participation in social security governance in an orderly and reasonable manner.

Research on Zero-failure Data Airborne Redundancy EWIS Reliability

XIAO Chu-wan1， DENG Li2， ZHANG Zhen3

2019, 0(04): 98. doi:10.3969/j.issn.1006-2475.2019.04.018

Asbtract ( 170 )

PDF (694KB) ( 95 )

References | Related Articles | Metrics

Considering the life span of high reliability redundancy EWIS components obeying exponential distribution with unknown parameters, a zero-failure data reliability analysis method was put forward to evaluating the reliability of EWIS. Monte-Carlo sampling method was applied to parallel-series EWIS life span, and the “minimization maximum” method was used to get the EWIS life span. Probability paper test was used to judge whether EWIS life obey Weibull distribution initially, and then Pearson fitness test was used to identify the distribution obeying Weibull distribution or not. Taken zero-failure data flight samples and the hypothesis of EWIS life obeying Weibull distribution into account, the zero-failure data reliability analysis method was applied to evaluate the reliability of EWIS. The research method is meaningful for analyzing redundancy designed EWIS reliability with zero-failure data.

Integrated Public Key Encryption and Public Key Encryption with Keyword Search

ZENG Qi, HAN Xiao, CAO Yong-ming

2019, 0(04): 103. doi:10.3969/j.issn.1006-2475.2019.04.019

Asbtract ( 275 )

PDF (678KB) ( 139 )

References | Related Articles | Metrics

Public key encryption with keyword search (PEKS) is a useful cryptographic primitive which allows one to delegate to an untrusted storage server the capability of searching on publicly encrypted data without impacting the security and privacy of original data. However, due to lack of data encryption and decryption function, a PEKS scheme cannot be used alone but has to be coupled with a standard public key encryption (PKE) scheme. For this reason, a new cryptographic primitive called integrated PKE and PEKS (PKE+PEKS) was introduced by Baek et al. in 2006, which provides the functions of both PKE and PEKS. So far, several PKE+PEKS schemes have been proposed in the literature. However, none of them considers the keyword guessing attack. This paper proposes a new efficient PKE+PEKS scheme which can resist keyword guessing attacks. Compared with other existed scheme, the performance of this scheme is greatly improved and we needn’t use bilinear pairing in the generation of ciphertext of keywords and data, which reduces the cost of computation and storage. The security analysis shows that the scheme proposed in this paper can satisfy the security of ciphertext privacy, the trapdoor indistinguishability and the keyword guessing attack respectively. Efficiency analysis shows that the proposed scheme is more efficient.

Research and Implementation of Beidou Short Message Security #br# Protection System Based on National Secret Algorithm

YANG Chu-hua1， ZHOU Hang-fan2， MA Jun2， FU Ning2

2019, 0(04): 108. doi:10.3969/j.issn.1006-2475.2019.04.020

Asbtract ( 229 )

PDF (1511KB) ( 120 )

References | Related Articles | Metrics

In view of the increasing application of domestic Beidou short message communication in power system and the release of power system security protection requirements, this paper takes the electricity information collection system as an example to analyze the application status and security risks of Beidou short message communication. Combined with the power system security protection requirements, the security protection scheme of the mining system for the Beidou short message communication based on the national secret algorithm is proposed, and the security information exchange of the business data of the Beidou short message mining system is realized, including two-way identity authentication, key agreement and data encryption. Finally, the experimental test analyzes and proves its effectiveness.

An Entropy Weight Fusion Localization Algorithm for #br# Terminal Intrusion Detection in Substation

JI Tian-ming1， ZHUANG Ling2， YU Jun1， ZHU Guang-xin1, WANG Zhao1， MIAO Jing-wen1

2019, 0(04): 114. doi:10.3969/j.issn.1006-2475.2019.04.021

Asbtract ( 123 )

PDF (1112KB) ( 107 )

References | Related Articles | Metrics

The use of intelligent robots and mobile operation terminals accelerates the speed of the realization of unattended substation. These wireless terminals need to be managed and monitored in a unified way. In particular, the illegal access terminal must be detected and screened. For the reason that the possibility of forging MAC address information of illegal terminal cannot be excluded. Location information is used as the terminal identity, so it can be detected and illegal terminals can be discriminated. At present, the distance calculation localization method and the database matching localization method have been proposed. In this paper, combining with the triangle distance calculation method and database matching method, a new entropy weight fusion localization algorithm is provided. The algorithm determines the entropy weight according to the concept of information entropy in the information theory. It improves the intelligent robot positioning accuracy and the ability to detect illegal terminal.

Information System Security Risk Assessment Based on Incomplete Information Game Model

MI Qian-kun1, WU Bin2, DU Ning1,QIN Xi1

2019, 0(04): 118. doi:10.3969/j.issn.1006-2475.2019.04.022

Asbtract ( 160 )

PDF (1482KB) ( 107 )

References | Related Articles | Metrics

Game theory has the characteristics of opposition target, non-cooperation relationship and dependence policy. It is consistent with the network attack and defense process. Applying game theory to network information security has become a research hotspot, but most of the existing research results have adopted complete information. The game model does not match the actual network attack and defense. Based on this, in order to improve the accuracy of information system risk assessment, this paper constructs a static Bayesian offensive and defensive game model under incomplete information conditions, and applies it to network information system security risk assessment to construct a corresponding information system security risk assessment algorithm. The effectiveness of the model and method is verified by experiments, which can provide a scientific and effective evaluation of information system security threats.

Table of Content