Computer and Modernization

A Heuristic Linear Regression Loss Function Selection Method

ZHANG Yi

2017, 0(8): 1. doi:10.3969/j.issn.1006-2475.2017.08.001

Asbtract ( 266 )

References | Related Articles | Metrics

Loss function is used to quantify information loss and false degree in regression analysis. This paper addresses heuristic loss function selection for linear regression. For a given noise density, there exists an optimal loss function under an asymptotic setting i.e. squared loss is optimal for Gaussian noise density. However, in reallife applications the noise density is always unknown and the training samples are finite. Robust statistics provides ways for selecting the loss function using statistical information about noise density, however robust statistics is based on asymptotic assumption and may not be well applied for finite sample data sets. For such practical problems, we try to utilize concept of Vapnik’s εinsensitive loss function. We propose a heuristic method for setting the value of ε as a function of samples and noise variance. Experimental comparisons for linear regression problems show that the proposed loss function performs more robustly performance and yields higher prediction accuracy compared with popular squared loss and Huber’s leastmodulus loss.

Interval Prediction of Gradual Fault Parameter Based on Fuzzy Granular RVM

QU Xiaoyan1, ZHANG Yongliang2, LYU Xiaofeng1, MA Ling1

2017, 0(8): 5. doi:10.3969/j.issn.1006-2475.2017.08.002

Asbtract ( 173 )

References | Related Articles | Metrics

To solve the shortages of Relevance Vector Machine (RVM) in fault prediction for largescale samples, such as the fall of prediction accuracy, low computation efficiency and so on, a method of Fuzzy Granular RVM (FGRVM) used for interval prediction of gradual fault parameters was proposed. Firstly, the original time series of parameters were granulated by fuzzy information granulation (FIG), and then the redundant data in the training set was reduced by a reduction approach based on the exponent similarity. Secondly, the Simple Particle Swarm Optimization (tsPSO) algorithm was improved by adaptive extremum disturbed and adaptive mutation, and the kernel width of RVM was selected automatically by the improved tsPSO (itsPSO) algorithm with the optimization target to minimize the error of 4fold cross validation. Finally, the changeable intervals of gradual fault parameters were predicted by the welltrained RVM. The prediction performance of the method was proved by simulations on three classic largescale, nonlinear time series with noisy data and Delta Exhaust Gas Temperature (DEGT) of an aeroengine in Air China, and the experiment results show that, compared with RVM and Fuzzy Granular Support Vector Machine (FGSVM), the prediction accuracy and computation efficiency are improved greatly by the method proposed in the paper.


An Optimum Routing Algorithm for Wireless Sensor Networks #br# Based on Ant Colony Optimization

SUN Zongxing， CAI Liping， LI Shibao

2017, 0(8): 13. doi:10.3969/j.issn.1006-2475.2017.08.003

Asbtract ( 178 )

References | Related Articles | Metrics

One of the research hotspots in Wireless Sensor Networks (WSNs) is the realization of efficient and realtime data routing with limit resources. We proposed an Optimum Routing Algorithm (ORA) based on ant colony optimization, aiming at the differences of applications with different design objectives. To meet the requirements of energy efficiency, load balance and realtime performance, we introduced a new parameter. Meanwhile the routing algorithm of WSNs is optimized by the combination of ant colony optimization algorithm. The simulation results show that ORA can extend network lifetime as long as possible for different applications, and perform better in realtime, efficient routing.


Mounting Sequence Optimization for Vertical Rotary Chip Mounter #br# Based on Improved Ant Colony Algorithm

FU Yongzhong, PAN Yunfeng

2017, 0(8): 17. doi:10.3969/j.issn.1006-2475.2017.08.004

Asbtract ( 229 )

References | Related Articles | Metrics

The mounting sequence optimization problem of the vertical rotary chip mounter is studied under the condition that the feeder allocations are confirmed. An improved ant colony algorithm is proposed based on the mathematical model. The mount process is divided into collect/posted cycles and each cycle connection. It also introduces elite ant and the pheromone updating strategy based on global and local combination. The experimental results show that the improved algorithm has better precision, global search ability and faster convergence rate. Compared with the basic ant colony algorithm, the efficiency of the optimal solution is increased by about 11%.


A Chaos Quantumbehaved Particle Swarm Optimization #br# Algorithm Based on Differential Evolution

WANG Zhi

2017, 0(8): 22. doi:10.3969/j.issn.1006-2475.2017.08.005

Asbtract ( 235 )

References | Related Articles | Metrics

In view of the shortcomings of the quantumbehaved particle swarm optimization algorithm that is prone to premature convergence to the local optimal solution, this

paper proposed a chaos quantumbehaved particle swarm algorithm based on differential evolution. In order to enhance the diversity of the particles, the chaotic sequences of

Logistic maps are introduced into the particle in this algorithm. Then, the crossover and mutation operation is performed for the premature particles which can avoid the local

optimum in the latter part of the particle. Experimental results show that the proposed algorithm has faster convergence speed and better convergence performance than quantum

particle swarm algorithm.


Population Forecast and Analysis Based on Two Improved Population #br# Models Under New Fertility Policy——Taking Xi’an as an Example

HE Jinyang1， QIU Hongchao1， LI Lin2， WANG Zhuang3

2017, 0(8): 26. doi:10.3969/j.issn.1006-2475.2017.08.006

Asbtract ( 211 )

References | Related Articles | Metrics

Precise forecasting of population is the important basis of enacting population policy and development strategy. Taking the influence of fertility pattern on

population development into consideration, this paper analyzed and improved two models of the GM (1,1) gray prediction model and Leslie population forecasting model. Taking Xian

as an example, two improved models were used to forecast the future population changes in different fertility models.By analyzing the conclusion, this paper also gave some

suggestions on enacting population policy and dealing with population problems for local government.


Design of Mobile AdHoc Network Router Under Complex Network

MA Anzhong

2017, 0(8): 31. doi:10.3969/j.issn.1006-2475.2017.08.007

Asbtract ( 182 )

References | Related Articles | Metrics

Traditional mobile ad hoc network router has disadvantage of outdated data encoding algorithm, low transmission rate, poor resistance to outside interference. For

these problems, a solution is put forward under the complex network of mobile ad hoc network router. AIST adaptive coding algorithm is used for mobile ad hoc network router data

encoding optimization to improve data sets. At the same time, filtering sensors are used for filtering interference signal to improve the overall antiinterference performance.

The design of bilateral gain of sending and receiving antenna is guaranteed to receive signals smoothly under the complex network environment for providing highquality mobile

Web experience to users. A set of management procedures is designed for user to change mobile network data parameters timely. Simulation test proves that the performance of the

mobile adhoc network router under complex network is excellent, can meet the requirements of daily use.


Using Motif to Characterize and Analyze Mesoscopic Features of Social Networks

YANG Zishu

2017, 0(8): 36. doi:10.3969/j.issn.1006-2475.2017.08.008

Asbtract ( 214 )

References | Related Articles | Metrics

Motif is an important mesoscopic structure which exists in networks. Motif discovery is a key problem in data analysis of social networks. By studying the existence of

motif in social networks topology structure， the evolution of networks can be understood better. The RandESU algorithm was used in motif detection and motif features

analysis. The difference of motifs features in social networks, power networks and Internet was compared and analyzed. The experimental result shows that social networks have

greater probabilities to form the structure of “triadic closure”. They have stronger community features and bigger clustering coefficients. And, nodes also have a tendency to

connect with the hubs nodes. We also find that although the number of nodes occupy “structural holes” is less, but it plays a key role in the information propagation. Through

research an important conclusions can be drawled: “preferential attachment” and “triadic closure” are the reasons for driving network evolution.


XML Parser in Chinese, Uyghur and Kazakh of Xinjiang Kunlun Digital Satellite Channel

WANG Liangliang1,2， YAN Wei3， ZHANG Jiawei1

2017, 0(8): 42. doi:10.3969/j.issn.1006-2475.2017.08.009

Asbtract ( 214 )

References | Related Articles | Metrics

This paper expounds how to realize the support of Uygur and Kazakh in Xinjiang Kunlun digital satellite channel. The data link is based on IP over DVBS technology,

oneway transmission mode. At the beginning of data broadcast, the XML file rules of Chinese, Uygur and Kazakh resources are designed. In the receiving terminal, the Python

language is used to realize the analysis of the XML in the Kunlun Linux operating system environment. The main functions of the parser are directory traversal, content reading,

data storage, fault tolerance mechanism and so on. The data integrity detection and the data analysis and storage are mainly introduced.


A Greenhouse Monitoring System Based on Wireless Sensor Network

XIAO Linglu1,2

2017, 0(8): 46. doi:10.3969/j.issn.1006-2475.2017.08.010

Asbtract ( 153 )

References | Related Articles | Metrics

In order to solve the problem of centralized monitoring of large greenhouse, an environment monitoring system based on ZigBee wireless sensor network is designed. The

embedded gateway is used as the core of the control system, and the environmental parameters are acquired by sensor nodes distributed in greenhouse. ZigBee network is used to

realize wireless transmission of data and instruction in monitoring area. The embedded Web server is adopted to realize remote monitoring of multiple greenhouses, according to

the monitoring results, the opening and closing of agricultural facilities can be controlled automatically or manually to adjust greenhouse environment timely. The system is

flexible, accurate and scalable, and it provides an effective solution for the centralized monitoring of greenhouse.


A Deinterlacing Scheme Based on Scene Change Detection

ZHU Xiaotao1, WANG Xiaoqi2, DU Xiaodan3, HUANG Qian1

2017, 0(8): 51. doi:10.3969/j.issn.1006-2475.2017.08.011

Asbtract ( 194 )

References | Related Articles | Metrics

This paper proposes a motioncompensated deinterlacing scheme based on the results of scene change detection. The design idea consists of five parts: scene change

detection, motion estimation, motion state analysis, motion consistency analysis and motion compensation interpolation. Firstly, absolute pixel differences between adjacent

pictures are compared to determine whether scene change exists. If scene change exists, only backward motion vector candidates are considered; otherwise, bidirectional

prediction is conducted. Secondly, motion analysis is utilized to determine whether the motion estimation should be performed again with a smaller block size. Finally, motion

compensated interpolation is performed. Experimental results show that the proposed scheme achieves better subjective and objective results compared with conventional

deinterlacing approaches.


Vehicle Detection Based on Deep Learning

WANG Deyu1， XU Youchun2， LI Yongle2， LU Feng1， ZHENG Kaiwen1

2017, 0(8): 56. doi:10.3969/j.issn.1006-2475.2017.08.012

Asbtract ( 305 )

References | Related Articles | Metrics

Aiming at the detection problem of the vehicle in the general detection environment for selfdriving cars, the vehicle detection is transformed into the object

detection in the image, and the object detection algorithm based on deep learning is applied to the detection of the vehicle. A lot of experiments have been carried out to

verify the validity of the detection system by building the image data set of the vehicle. The detection system is an end to end detection system, only needs to input images,

which can quickly and accurately locate the position of the vehicle in the image and predict class probabilities.


Multithreshold Segmentation of Color Image Based on Hybrid PSOCS Algorithm

WEI Hongchun

2017, 0(8): 61. doi:10.3969/j.issn.1006-2475.2017.08.013

Asbtract ( 172 )

References | Related Articles | Metrics

Color image segmentation is a difficult problem in digital image processing. In this paper, the use of swarm intelligence algorithm for color image segmentation is studied. For the defect on the cuckoo algorithm caused by optimization of Levy flight jumping, after each end of Levi flight, a modified particle swarm location variation equation is introduced to guide the global optimal solution, and the corresponding adaptive equations are also proposed for the discovery probability and the pace factor. So a hybrid particle swarm cuckoo algorithm(HPCS) is proposed, which is used to carry on multithreshold segmentation of color image. The experiment shows that the HPCS algorithm is good in the efficiency and quality of color image segmentation.

Boundary Sample Selection Method Based on Cosine Similarity

LI Chunli， LIU Zhendong， HUI Kanghua

2017, 0(8): 66. doi:10.3969/j.issn.1006-2475.2017.08.014

Asbtract ( 293 )

References | Related Articles | Metrics

Abstract: The training of convolution neural network usually requires a lot of training samples, which causes the training time be too long. To solve this problem, this paper presents a boundary sample selection method based on cosine similarity. We select boundary samples as the training set of convolution neural network, and carry out example selection experiment on the MNIST, CIFAR10 and SVHN data sets. Then a convolutional neural network is used to carry out experiments. Experimental results show that this method can preserve the typical samples in the training set and eliminate redundant samples. Thereby, the number of training samples is reduced, the network training time is shortened and the learning efficiency of network is improved.

Application of Information Visualization in Monitoring #br# and Optimization for Wireless Internet

DENG Xianyu1, GUO Daming2

2017, 0(8): 71. doi:10.3969/j.issn.1006-2475.2017.08.015

Asbtract ( 172 )

References | Related Articles | Metrics

Wireless Internet is developing rapidly recently, and people’s requirements for wireless network are also rising at the same time, so the monitoring and optimization work of wireless network is very important. The traditional monitoring and optimization interface shows the wireless network parameter information and user complaints in tabular form, while this paper aims to use information visualization technology combined with the characteristics of wireless networks to study and design its visualization program, so as to efficiently manage the wireless network.


3D Reconstruction of FMAV Based on MATLAB

LOU Xingliang, ZHANG Weiping, CAI Xuefei, ZOU Yang

2017, 0(8): 74. doi:10.3969/j.issn.1006-2475.2017.08.016

Asbtract ( 248 )

References | Related Articles | Metrics

FMAV usually works at high frequency, thus its wing is usually hard to be observed. This paper offers a new way to solve this problem. Two highspeed cameras are used to record the motion of the wing and the data are calculated through 3D reconstruction algorithm. By this way we can get the spatial coordinates of certain points on the wing. This method is verified by comparing the experimental data with theory data.

An Embedded Software Reliability Modeling and Evaluation Method

LIU Weiwei, ZHUANG Yi, LI Mi

2017, 0(8): 78. doi:10.3969/j.issn.1006-2475.2017.08.017

Asbtract ( 210 )

References | Related Articles | Metrics

Although AADL has been widely used in the modeling and analysis of embedded software architecture, it’s a semiformal modeling language and cannot meet the

requirements of rigorous analysis and evaluation of software reliability. In order to solve this problem, first of all, we describe the AADL reliability model by discretetime

Markov chain, which is mainly used to describe the state transition and transition probability of the system. Then, we propose a reliability quantitative evaluation method based

on PCTL to evaluate the reliability by calculating the availability. Finally, we design a reliability evaluation tool. By the tool, a case study of a flight management system is

given to verify the effectiveness of the modeling and evaluation method.


Ontologybased Consistency Verification of Conceptual Architecture Derivation

JIANG Dongyu1, KANG Dazhou1,2, WANG Shun1

2017, 0(8): 84. doi:10.3969/j.issn.1006-2475.2017.08.018

Asbtract ( 186 )

References | Related Articles | Metrics

In the software product line engineering, the feature model is used to automatically derive the conceptual architecture. However, it is the main factor to reduce the

export efficiency whether manual detection concept architecture is consistent with the demand expression or not. In order to assist designers to carry out efficient conceptual

architecture derivation, this paper proposed an ontologybased method for consistency verification. Firstly, the paper modeled the conceptual architecture metamodel. On the

basis of this, the conceptual architecture ontology model was constructed and the corresponding reasoning rules were proposed. Then in the process of deriving the conceptual

architecture, the ontology was used to describe it. After that the SWRL language is used to convert the inference rules of the natural language description into the data format

which can be recognized by the ontology reasoning machine. Finally, the rule file and ontology description file were imported into the Jena inference engine to verify the

consistency. In this paper, a case of VOIP system is described in detail, which verifies the feasibility of the method, and compares the advantages and disadvantages of the

method in theory with other common consistency verification methods.

A Text Information Hiding Algorithm of Song Poetry Carrier #br# Based on Sentence Pattern Control

QU Qifeng， LIU Yanchen， WANG Jian， WANG Tianxing， YU Shun

2017, 0(8): 91. doi:10.3969/j.issn.1006-2475.2017.08.019

Asbtract ( 185 )

References | Related Articles | Metrics

In this paper, a text information hiding scheme of Song poetry carrier based on sentence pattern control is proposed for the problem that current text information

hiding algorithm is featured with low embedding rate, weak stealthiness and easy to be tested by machine. Through analyzing the relationship between rhythm and word class in

template of Song poetry versification, wordclass extraction algorithm of Song poetry dictionary was designed, capacity of “rhythmword class” dictionary generated after

processing exceeded 40000 words, with 4 word classes and 32 rhythms. Information hiding and extraction algorithm under the control of the designed sentence pattern could covert

the secret information in the form of 140 tune patterns in the complete Song poetry poems, to generate converted Song poetry with versification, sentence pattern and word class

completely same as the original Song poetry. The experiment result indicated that the success rate of word segmentation of this scheme was up to 100%, and embedding rate of

secret information to a maximum of 20.2%, characterized by high running speed, strong stealthiness and high safety when compared with Song poetry dictionary constructing

algorithm based on statistics.

Browser Cache Collaboration System Based on P2P

GAO Fei

2017, 0(8): 98. doi:10.3969/j.issn.1006-2475.2017.08.020

Asbtract ( 157 )

References | Related Articles | Metrics

A browser cache collaboration system based on hybrid P2P network was proposed. Its system structure was expounded. And the design and implementation of the various

modules of the system were analyzed. The core module of the system—index was improved. High efficient index method suitable for the application was proposed. Experiments showed

that the efficiency of the improved index was higher, which can satisfy the requirements of the application. The whole system was feasible. For the overall performance, the

system can remarkably shorten the response time than direct access to the Internet and the response time almost had no relationship with the file size.

A Social Psychology Trust Model for P2P Network

REN Mengchen1, LIU Guangjie1, LIU Weiwei1, ZHAI Jiangtao2, DAI Yuewei1,2

2017, 0(8): 104. doi:10.3969/j.issn.1006-2475.2017.08.021

Asbtract ( 192 )

References | Related Articles | Metrics

To achieve dynamic adaptability and low computational complexity for PeertoPeer (P2P) network trust model, a trust model based on social psychology for P2P Network

is proposed, which draws on the three basic concepts of proximity, familiarity and similarity in social psychology. Mature trading relationship, trust learning factor,

transaction density factor and similarity factor are defined to obtain direct trust value and recommendation trust value, and further to calculate resulting trust value. The

proposed model is verified effective with simulation results on dynamic adaptability, fault tolerance and superiority.

Alarm Information Collection Method in Railway Communication #br# Transmission System Based on CORBA

LI Hongren

2017, 0(8): 109. doi: 10.3969/j.issn.1006-2475.2017.08.022

Asbtract ( 105 )

References | Related Articles | Metrics

In order to solve the problem that the equipment standards and types in the railway transmission network are complicated, and the interface between the different

vendors is not compatible, or the management and maintenance problems caused by the data is not open, in this paper, we deeply analyze the current developing situation of the

railway transmission network, make a study of northbound CORBA interface characteristics and properties, and put forward a method of alarm information collection in railway

communication transmission system based on CORBA combined with the actual problem. The application in real railway shows that the method can collect the equipment alarm

information from different manufacturer realtimely and can improve the level and efficiency of the network operation maintenance.


Energyconscious Behavior Analysis of Mobile Phones

CHENG Rujie1， LU Jian1， JIANG Houming2， HU Mu2， WU Jia3

2017, 0(8): 114. doi:10.3969/j.issn.1006-2475.2017.08.023

Asbtract ( 197 )

References | Related Articles | Metrics

Energyconscious behavior analysis plays a vital part in energy consumption analysis of mobile phones, it is of great help to energy optimization by understanding how

and why the energy is used. This paper presents a method which combined software detection and hardware detection on both iOS and Android platforms. In terms of software

detection, the battery level was obtained based on system APIs and relevant applications which were developed to measure the energy consumption of key energyconsuming

entities, such as screen, audio and network interface under different using scenarios， so that the key factors could be determined. In terms of hardware detection, NI myDAQ was

used to communicate with computer. By recording the current in real time, the energy consumption was calculated. The experiment result of hardware detection validates the

effectiveness of software detection. In the end, by comparing all the energyconscious behaviors, the highly conscious behaviors are identified.

A University’s Answering System Based on WeChat and Java Web

YANG Peiji

2017, 0(8): 123. doi:10.3969/j.issn.1006-2475.2017.08.024

Asbtract ( 221 )

References | Related Articles | Metrics

To get rid of problems such as the limitation of time and space in the process of teaching and answering in colleges, an online answering system is developed based on

WeChat and Java Web with the aim to improve the quality of teaching in colleges. The system enables teachers to get control over curriculums, students and questions asked by

students through WeChat public platform. With the help of WeChat, students can easily achieve the answer. The system is able to effectively keep the costs of the Q&A session as

low as possible so that the quality of Q&A will be improved.


Table of Content