����Ŀ¼

15 August 2020, Volume 6 Issue 4

Previous Issue    Next Issue

Comprehensive Review

Survey of text classification methods based on deep learning

Sijia DU,Haining YU,Hongli ZHANG

2020, 6(4):  1-13.  doi:10.11959/j.issn.2096-109x.2020010

Asbtract ( 970 )   HTML ( 173)   PDF (1259KB) ( 758 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Text classification is a research hot spot in the field of natural language processing,which is mainly used in public opinion detection,news classification and other fields.In recent years,artificial neural networks has good performance in many tasks of natural language processing,the application of neural network technology to text classification has also made many achievements.In the field of text classification based on deep learning,numerical representation of text and deep-learning-based text classification are two main research directions.The important technology of word embedding in text representation and the implementation principle and research status of deep learning method applied in text classification were systematically analyzed and summarized.And the shortcomings and the development trend of text classification methods in view of the current technology development were analyzed.

Special Column: Application Exploration on Multi-party Trusted Cooperation based on Blockchain

Lightweight mobile Ad Hoc network authentication scheme based on blockchain

Xu ZHANG,Xin MA

2020, 6(4):  14-22.  doi:10.11959/j.issn.2096-109x.2020040

Asbtract ( 456 )   HTML ( 65)   PDF (1158KB) ( 727 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to solve the problem that the existing authentication method can’t track the vulnerable of malicious nodes and depends on the centralized authentication center,based on the de-centralization,tamper-proof and traceable characteristics of blockchain technology,a distributed lightweight mobile Ad Hoc network authentication scheme was proposed.It included the establishment of peer-to-peer network,the design of certification ledger and the choice of consensus mechanism.Then the role of smart contracts was analyzed and discussed.In this scenario,when the maximum transmission delay τ_max between a pair of peers is given,the relationship between node authentication timeδand network size $N_t^{\text{all}}$is satisfied with$\left(\frac{N_t^{\text{all}}}{3}+7\right){\tau }_{\text{max}}\le \delta \le 3(N_t^{\text{all}}+1){\tau }_{\text{max}}$ .

Efficient and safe auditable mixed-coin service scheme based on blind signature

Kang QIAO,Hongbo TANG,Wei YOU,Haitao LI

2020, 6(4):  23-36.  doi:10.11959/j.issn.2096-109x.2020043

Asbtract ( 365 )   HTML ( 30)   PDF (2045KB) ( 730 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The mixed-coin service can provide solutions for the privacy problem of blockchain,but it still faces efficiency bottlenecks and security risks.To further improve the efficiency and security protection of the mixed-coin service,an efficient and safe auditable mixed-coin service scheme based on blind signature was proposed.Firstly,this scheme added audit measures.It added an audit blockchain to the traditional mix-coin model to record the behavior of users and mixers,achieving traceability and accountability.Then,this method used elliptic curves algorithm to construct blind signatures instead of blind signature schemes based on bilinear pairs or RSA.Finally,this scheme proposed an auditable blind mix-coin service agreement based on the auditable mix-coin model and the blind signature algorithm based on elliptic curve.Simulation analysis shows that the proposed scheme has six security features,such as auditability and anti-theft attack,while providing privacy protection.Under the same security intensity,the proposed scheme can effectively reduce the computational overhead and storage overhead.

PKI cross-domain authentication scheme based on advanced PBFT algorithm

Sijie QIAN,Liquan CHEN,Shihui WANG

2020, 6(4):  37-44.  doi:10.11959/j.issn.2096-109x.2020042

Asbtract ( 410 )   HTML ( 60)   PDF (974KB) ( 929 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to solve the efficiency problem of the existing public key infrastructure cross-domain authentication scheme,a cross-domain authentication model based on the consortium blockchain which has the advantages of distributed and difficult to be tamperd with was proposed.On the one hand,the dynamic join and exit function was added to the practical Byzantine fault tolerant (PBFT) algorithm,the primary node election mode was improved,and the three-stage broadcast was reduced to two-stage broadcast for the reducation of communication overhead.On the other hand,the cross-domain authentication system architecture based on consortium chain was designed,the blockchain certificate format and the cross-domain authentication protocol were presented,the security and efficiency were analyzed.The results shows that in term of security,the proposed model has security attributes such as resisting distributed attacks.In terms of performance,the proposed model has advantages in both computational overhead and communication overhead when it was compared with the existing cross-domain authentication schemes.

Data sharing scheme for 5G IoT based on blockchain

Kang QIAO,Wei YOU,Lingwei WANG,Hongbo TANG

2020, 6(4):  45-55.  doi:10.11959/j.issn.2096-109x.2020041

Asbtract ( 657 )   HTML ( 111)   PDF (3659KB) ( 909 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The massive amount of IoT data has great value,but the existing cloud-based data sharing mechanism faces problems such as single points of failure and internal leakage,which can not ensure the secure sharing of user data.In order to achieve efficient and reliable data sharing,the blockchain technology was used to propose a 5G IoT data sharing scheme based on blockchain.Firstly,the proposed scheme designed data sharing framework and data sharing process,based on the lightning network solution,an off-chain transaction mechanism for data sharing of the internet of things was proposed.Experimental analysis shows that the 5G IoT data sharing framework based on blockchain has strong anti-attack ability and the off-chain transaction mechanism based on lightning network can greatly increase transaction throughput and reduce transaction delay.

Multi-microgrid marketized transaction models and algorithms based on smart contracts

Fen LI,Jin LI,Songqi WU,Senlin ZHANG,Yueming LU

2020, 6(4):  56-66.  doi:10.11959/j.issn.2096-109x.2020023

Asbtract ( 299 )   HTML ( 49)   PDF (1985KB) ( 490 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Facing the problems of high cost,low efficiency and low transparency existing in the traditional transmission and distribution networks under the centralized management,and the demand for maximizing the interests of all players in the multi-microgrid system,a fair bidding rule and multi-objective dynamic planning model were established.Using blockchain trusted ledger and smart contract technology,a decentralized,transparent,open and information symmetrical market-oriented transaction model was established to ensure the interests of all parties.The simulation results show that the model allocation strategy can effectively balance the interests of market participants,and verify the applicability and effectiveness of the bidding rules in multi microgrid market-oriented transaction.

Papers

Improve the robustness of algorithm under adversarial environment by moving target defense

Kang HE,Yuefei ZHU,Long LIU,Bin LU,Bin LIU

2020, 6(4):  67-76.  doi:10.11959/j.issn.2096-109x.2020052

Asbtract ( 256 )   HTML ( 30)   PDF (1062KB) ( 364 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Traditional machine learning models works in peace environment,assuming that training data and test data share the same distribution.However,the hypothesis does not hold in areas like malicious document detection.The enemy attacks the classification algorithm by modifying the test samples so that the well-constructed malicious samples can escape the detection by machine learning models.To improve the security of machine learning algorithms,moving target defense (MTD) based method was proposed to enhance the robustness.Experimental results show that the proposed method could effectively resist the evasion attack to detection algorithm by dynamic transformation in the stages of algorithm model,feature selection and result output.

MLAR:large-scale network alias resolution for IP geolocation

Fuxiang YUAN,Fenlin LIU,Chong LIU,Yan LIU,Xiangyang LUO

2020, 6(4):  77-94.  doi:10.11959/j.issn.2096-109x.2020044

Asbtract ( 345 )   HTML ( 36)   PDF (4290KB) ( 314 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to accurately and efficiently perform alias resolution on interface IP and support IP geolocation,a large-scale network alias resolution algorithm (MLAR) was proposed.Based on the statistical differences in delays,paths,Whois,etc.between alias IP and non-alias IP,before resolution,filtering rules were designed to exclude a large number of IPs that can not be aliases and improve efficiency of resolution,alias resolution was transformed into classification,and four novel features such as delay similarity,path similarity,etc.were constructed for the classification of possible alias IP and non-alias IP after filtering.Experiments based on millions of samples from CAIDA show that compared with RadarGun,MIDAR,and TreeNET,the accuracy is improved by 15.8%,4.8%,5.7%,the time consumption can be reduced by up to 77.8%,65.3%,and 55.2%,when the proposed algorithm is applied to IP geolocation,the failure rates of the three typical geolocation methods such as SLG,LENCR,and PoPG are reduced by about 65.5%,64.1%,and 58.1%.

Function argument number identification in stripped binary under PPC and MIPS instruction set

Xiaokang YIN,Liu LIU,Long LIU,Shengli LIU

2020, 6(4):  95-103.  doi:10.11959/j.issn.2096-109x.2020047

Asbtract ( 281 )   HTML ( 23)   PDF (1029KB) ( 347 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The identification of the number of function argument contributes to the recovery of the function prototype and is the basis for data flow analysis and other security analysis.In order to improve the accuracy of the recognition of the number of function parameters,an algorithm (Findargs) which determines the number of parameters of the function according to the voting mechanism of the function call relationship was proposed.Findargs starts from the function call characteristics of PPC and MIPS instruction set,and uses function call relationship combined with argument pass analysis to identify the number of function arguments,which can help to recover function prototype.In order to evaluate the recognition effect of Findargs,a large binary file was selected and tested it with radare2.The experiments results show that Findargs has higher accuracy,and the accuracy rate for PPC instruction set reaches 90.3%.For MIPS instruction set,the accuracy rate is 86%.

De-anonymiation method for networks based on DeepLink

Pei WANG,Yan JIA,Aiping LI,Qianyue JIANG

2020, 6(4):  104-108.  doi:10.11959/j.issn.2096-109x.2020045

Asbtract ( 306 )   HTML ( 27)   PDF (790KB) ( 278 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Existing de-anonymization technologies are mainly based on the network structure.To learn and express network structure is the key step of de-anonymization.The purpose of the user identity linkage is to detect the same user from different social networking platforms.DeepLink is a cross-social network user alignment technology.It learns the structural of the social networks and align anchor nodes through deep neural networks.DeepLink was used to identify de-anonymization social networks,and the results outperforms the traditional methods.

Differentially private sequence generative adversarial networks for data privacy masking

Yu ZHANG,Xixiang LYU,Yucong ZOU,Yige LI

2020, 6(4):  109-119.  doi:10.11959/j.issn.2096-109x.2020046

Asbtract ( 476 )   HTML ( 44)   PDF (1593KB) ( 710 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Based on generative adversary networks and the differential privacy mechanism,a differentially private sequence generative adversarial net (DP-SeqGAN) was proposed,with which the privacy of text sequence data sets can be filtered out.DP-SeqGAN can be used to automatically extract important features of a data set and then generate a new data set which was close to the original one in terms of data distributions.Based on differential privacy,randomness is introduced to the model,which improves the privacy of the generated data set and further reduces the over fitting of the discriminator.The proposed DP-SeqGAN was universal,so there is no need to adjust the model adaptively for datasets or design complex masking rules against dataset characters.The experiments show that the privacy and usability of a sequence data set are both improved significantly after it is processed by the DP-SeqGAN model,and DP-SeqGAN can greatly reduce the success rate of member inference attacks against the generated data set.

Improvement of lightweight threshold group signature scheme

Zhirong ZHU,Jiayu LYU,Guohua YE,Zhiqiang YAO

2020, 6(4):  120-129.  doi:10.11959/j.issn.2096-109x.2020048

Asbtract ( 219 )   HTML ( 24)   PDF (907KB) ( 267 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to improve the security and efficiency of threshold group signature in 5G environment,a new scheme was proposed.The core idea was to introduce message token and dynamic identity to improve the freshness of authentication information and increase the difficulty of enemy attacks.At the same time,the number of interactions between the participating parties was simplified,and some computing tasks were migrated to the server to reduce the computing overhead of the member part.After security analysis,the proposed scheme can resist multiple attacks including replay attack.After performance analysis,the proposed scheme can reduce the interaction with the trusted center twice compared with similar schemes,reduce the signature space by 30%,and have shorter execution time,which is suitable for resource-constrained devices under 5G network.

Design method of secure computing protocol for deep neural network

Renwan BI,Qianxin CHEN,Jinbo XIONG,Ximeng LIU

2020, 6(4):  130-139.  doi:10.11959/j.issn.2096-109x.2020050

Asbtract ( 370 )   HTML ( 56)   PDF (1244KB) ( 396 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Aiming at the information leakage problem in the process of deep neural network model calculation,a series of secure and efficient interactive computing protocols were designed between two non-collusive edge servers in combination with the additive secret sharing scheme.Since the nonlinear function cannot be split directly,a set of basic conversion protocols were proposed to realize the secure conversion of additive and multiplicative shares.After a few invokes,the power,comparison,exponential,logarithm,division and other low-level functions can be calculated securely.Due to the characteristics of data transfer and computation,the proposed protocols can be extended to array computation.Theoretical analysis ensures the correctness,efficiency and security of these protocols.The experimental results show that the error of these protocols is negligible,and the computational costs and communication overhead are better than the existing schemes.

Analysis and design of visual secret sharing scheme for gray image

Jian LIU,Yi ZHANG,Yuling LIU,Hao HU

2020, 6(4):  140-147.  doi:10.11959/j.issn.2096-109x.2020049

Asbtract ( 219 )   HTML ( 19)   PDF (6653KB) ( 405 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Existing gray visual secret sharing (GVSS) convert the original image into black-white binary image for processing,which causes the pixel information loss problem.For this purpose,a kind of GVSS based on gray superposition was proposed.For each gray-scale pixel,the corresponding encoding matrix was designed to generate gray share image.On this basis,the schemes for (2,2),(2,n),(n,n) threshold structures were designed.Compared with the existing schemes,the proposed scheme effectively reduces pixel expansion and achieves the perfect reconstruction of gray secret images.

Quantum dialogue protocol based on self-dual quantum low density parity check codes

Fen LIU,Zhenxing HE,Pengao XU,Hongyang MA

2020, 6(4):  148-152.  doi:10.11959/j.issn.2096-109x.2020051

Asbtract ( 165 )   HTML ( 15)   PDF (1238KB) ( 268 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In quantum channels,particles are usually affected by noise during transmission.A quantum dialogue protocol based on self-dual quantum low density parity check codes was proposed to resist noise attacks.A combination of B construction method and U construction method was used to construct a self-dual quantum low density parity check matrix.The proposed quantum dialogue protocol can resist common external attacks,and there was no information leakage,which improved the efficiency of encoding and decoding.The security of quantum dialogue protocol was studied from the perspective of error correction.Security analysis shows that the protocol has sufficient security and can effectively resist common malicious attacks.

Investigation,analysis and improvement mentality of the revision on information security’s training program to undergraduates in Hunan University

Qiao HU,Yupeng HU,Xin LIAO,Zheng QIN,Jianhua SUN,Jiliang ZHANG

2020, 6(4):  153-158.  doi:10.11959/j.issn.2096-109x.2020053

Asbtract ( 235 )   HTML ( 31)   PDF (16938KB) ( 143 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Based on the investigation result on the training program of information security to undergraduates in Hunan university,an analysis from four aspects was made:domestic and international universities,external specialists,teachers in the college and students in the college.Aiming at issues being found,the training program based on existing experiences and achievements on university-enterprise cooperation were improved.The new training program features university-enterprise joint cultivation.