Table of Content

10 October 2023, Volume 43 Issue 10

Previous Issue   

Blockchain technology

Overview of cryptocurrency regulatory technologies research

Jiaxin WANG, Jiaqi YAN, Qian’ang MAO

2023, 43(10):  2983-2995.  DOI: 10.11772/j.issn.1001-9081.2022111694

Asbtract ( )   HTML ( )   PDF (911KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the help of blockchain and other emerging technologies， cryptocurrencies are decentralized， autonomous and cross-border. Research on cryptocurrency regulatory technologies is not only helpful to fight criminal activities based on cryptocurrencies， but also helpful to provide feasible supervision schemes for the expansion of blockchain technologies in other fields. Firstly， based on the application characteristics of cryptocurrency， the Generation， Exchange and Circulation （GEC） cycle theory of cryptocurrency was defined and elaborated. Then， the frequent international and domestic crimes based on cryptocurrencies were analyzed in detail， and the research status of cryptocurrency security supervision technologies in all three stages was investigated and surveyed as key point. Finally， the cryptocurrency regulatory platform ecology systems and current challenges faced by the regulatory technologies were summarized， and the future research directions of cryptocurrency regulatory technologies were prospected in order to provide reference for subsequent research.

Survey on privacy-preserving technology for blockchain transaction

Qingqing XIE, Nianmin YANG, Xia FENG

2023, 43(10):  2996-3007.  DOI: 10.11772/j.issn.1001-9081.2022101555

Asbtract ( )   HTML ( )   PDF (2911KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Blockchain ledger is open and transparent. Some attackers can obtain sensitive information through analyzing the ledger data. It causes a great threat to users’ privacy preservation of transaction. In view of the importance of blockchain transaction privacy preservation， the causes of the transaction privacy leakage were analyzed at first， and the transaction privacy was divided into two types： the transaction participator’s identity privacy and transaction data privacy. Then， in the perspectives of these two types of transaction privacy， the existing privacy-preserving technologies for blockchain transaction were presented. Next， in view of the contradiction between the transaction identity privacy preservation and supervision， transaction identity privacy preservation schemes considering supervision were introduced. Finally， the future research directions of the privacy-preserving technologies for blockchain transaction were summarized and prospected.

Survey on combination of computation offloading and blockchain in internet of things

Rui MEN, Shujia FAN, Axida SHAN, Shaoyu DU, Xiumei FAN

2023, 43(10):  3008-3016.  DOI: 10.11772/j.issn.1001-9081.2022091466

Asbtract ( )   HTML ( )   PDF (882KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the recent development of mobile communication technology and the popularization of smart devices， the computation-intensive tasks of the terminal devices can be offloaded to edge servers to solve the problem of insufficient resources. However， the distributed nature of computation offloading technology exposes terminal devices and edge servers to security risks. And， blockchain technology can provide a safe environment transaction for the computation offloading system. The combination of the above two technologies can solve the insufficient resource and the security problems in internet of things. Therefore， the research results of applications combining computation offloading with blockchain technologies in internet of things were surveyed. Firstly， the application scenarios and system functions in the combination of computation offloading and blockchain technologies were analyzed. Then， the main problems solved by blockchain technology and the key techniques used in this technology were summarized in the computation offloading system. The formulation methods， optimization objectives and optimization algorithms of computation offloading strategies in the blockchain system were classified. Finally， the problems in the combination were provided， and the future directions of development in this area were prospected.

Cross-chain review： mechanisms， protocols， applications and challenges

Longfei CHEN, Zhongyuan YAO, Heng PAN, Gaoyuan QUAN, Xueming SI

2023, 43(10):  3017-3027.  DOI: 10.11772/j.issn.1001-9081.2022111643

Asbtract ( )   HTML ( )   PDF (1641KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the continuous development of the blockchain technology and its applications， the demand for interoperability among blockchains is increasing. However， the isolation and closeness of blockchain as well as the heterogeneity among different blockchains cause the "island of value" effect of blockchain， which seriously hinder the widespread implementation and sound development of blockchain based integrated applications. Cross-chain technology of blockchain solves the problems of data circulation， value transfer and business collaboration among different blockchains， and is also an important approach to improve the scalability and interoperability of blockchains. According to the degrees of the implementation complexity and the function richness of cross-chain technology， the cross-chain technology of blockchain was summarized and then classified into three types： the basic cross-chain mechanisms， the cross-chain protocols based on these mechanisms， and the cross-chain applications with system architectures. Finally， the existing problems in cross-chain interoperations were summed up， thereby providing systematical and theoretical reference for the further research on cross-chain technology of blockchain.

Cross-chain privacy protection scheme of consortium blockchain based on improved notary mechanism

Xiaohan GUO, Zhongyuan YAO, Yong ZHANG, Shangkun GUO, Chao WANG, Xueming SI

2023, 43(10):  3028-3037.  DOI: 10.11772/j.issn.1001-9081.2022111641

Asbtract ( )   HTML ( )   PDF (3640KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Cross-chain interaction of consortium blockchain not only enhances the function of the application of consortium blockchain， but also expands the scope of usage of the application， so that it is of great significance to the application promotion and industrial development of consortium blockchain. However， the cross-chain interaction of consortium blockchain still has the privacy disclosure problems of user identity and asset transaction information at present， which has become a major factor hindering the wide application of the cross-chain interaction technology of consortium blockchain. In view of the above problems， a cross-chain privacy protection scheme for consortium blockchain assets based on improved notary mechanism was proposed. In the scheme， a hash locking mechanism was introduced at the contract layer to improve the traditional single-signature notary cross-chain method at first， so as to reduce the risk of the traditional notary mechanism centralizing and doing evil. Then， the characteristics of homomorphic encryption were used to realize the usability and invisibility of transaction assets under the premise of ensuring the legitimacy of transactions. At the same time， the identity-based cryptographic algorithm of multi-Key Generation Center （KGC） mode was used to protect the user identity privacy at the network layer. The theoretical analysis and experimental results show that the proposed scheme has a good privacy protection effect on the user identity information and asset information in the cross-chain interaction of consortium blockchain， and this scheme has lower overhead in signature and verification than other similar schemes.

Survey of visualization research based on blockchain technology and application

Yimin SHAO, Fan ZHAO, Yi WANG, Baoquan WANG

2023, 43(10):  3038-3046.  DOI: 10.11772/j.issn.1001-9081.2022111642

Asbtract ( )   HTML ( )   PDF (8589KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Blockchain technology， which is originated from Bitcoin， is a disruptive and innovative technology with very broad development prospects. Facing the expansion of demand of blockchain platforms and application fields， the introduction of visualization technology can enhance users’cognitive ability， help users efficiently discover useful information from massive and complex data， and facilitate users’understanding and decision-making， which is one of the frontiers of blockchain technology research. In order to gain a deeper understanding of the visualization research based on blockchain technology and application， firstly， the basic theory of blockchain and visualization was introduced， and the existing literature on blockchain visualization was analyzed form multiple dimensions. Next， starting from the common key technologies， the visualization research methods in blockchain transaction processing， consensus mechanism， smart contract and network security were introduced. At the same time， the application status of blockchain visualization in various fields such as virtual currency， social livelihood and integrated innovation was outlined. Finally， the development trends of visualization research based on blockchain technology and application were summarized and prospected.

Artificial intelligence

Graph summarization algorithm based on node similarity grouping and graph compression

Yu HONG, Hongchang CHEN, Jianpeng ZHANG, Ruiyang HUANG

2023, 43(10):  3047-3053.  DOI: 10.11772/j.issn.1001-9081.2022101535

Asbtract ( )   HTML ( )   PDF (1105KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To solve the problem that the current graph summarization methods have high compression ratios and the graph compression algorithms cannot be directly used in downstream tasks， a fusion algorithm of graph summarization and graph compression was proposed， which called Graph Summarization algorithm based on Node Similarity grouping and graph Compression （GSNSC）. Firstly， the nodes were initialized as super nodes， and the super nodes were grouped according to the similarity. Secondly， the super nodes of each group were merged until the specified number of times or nodes were reached. Thirdly， super edges and corrected edges were added between the super nodes for reconstructing the original graph. Finally， for the graph compression part， the cost of compressing and summarizing the adjacent edges of each super node were judged， and the less expensive one in these two was selected to execute. Experiments of graph compression ratio and graph query were conducted on six datasets such as Web-NotreDame， Web-Google and Web-Berkstan. Experimental results on six datasets show that， the proposed algorithm has the compression ratio reduced by at least 23 percentage points compared with SLUGGER （Scalable Lossless sUmmarization of Graphs with HiERarchy） algorithm， and the compression ratio decreased by at least 13 percentage points compared with SWeG （Summarization of Web-scale Graphs） algorithm. Experimental results on Web-NotreDame dataset show that the degree error of the proposed algorithm is reduced by 41.6% compared with that of SWeG algorithm. The above verifies that the proposed algorithm has better graph compression ratio and graph query accuracy.

Network representation learning based on autoencoder with optimized graph structure

Kun FU, Yuhan HAO, Minglei SUN, Yinghua LIU

2023, 43(10):  3054-3061.  DOI: 10.11772/j.issn.1001-9081.2022101494

Asbtract ( )   HTML ( )   PDF (2515KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The aim of Network Representation Learning （NRL） is to learn the potential and low-dimensional representation of network vertices， and the obtained representation is applied for downstream network analysis tasks. The existing NRL algorithms using autoencoder extract information about node attributes insufficiently and are easy to generate information bias， which affects the learning effect. Aiming at these problems， a Network Representation learning model based on Autoencoder with optimized Graph Structure （NR-AGS） was proposed to improve the accuracy by optimizing the graph structure. Firstly， the structure and attribute information were fused to generate the joint transition matrix， thereby forming the high-dimensional representation. Secondly， the low-dimensional embedded representation was learnt by an autoencoder. Finally， the deep embedded clustering algorithm was introduced during learning to form a self-supervision mechanism in the processes of autoencoder training and the category distribution division of nodes. At the same time， the improved Maximum Mean Discrepancy （MMD） algorithm was used to reduce the gap between distribution of the learnt low-dimensional embedded representation and distribution of the original data. Besides， in the proposed model， the reconstruction loss of the autoencoder， the deep embedded clustering loss and the improved MMD loss were used to optimize the network jointly. NR-AGS was applied to the learning of three real datasets， and the obtained low-dimensional representation was used for downstream tasks such as node classification and node clustering. Experimental results show that compared with the deep graph representation model DNGR （Deep Neural networks for Graph Representations）， NR-AGS improves the Micro-F1 score by 7.2， 13.5 and 8.2 percentage points at least and respectively on Cora， Citeseer and Wiki datasets. It can be seen that NR-AGS can improve the learning effect of NRL effectively.

Sentence embedding optimization based on manifold learning

Mingyue WU, Dong ZHOU, Wenyu ZHAO, Wei QU

2023, 43(10):  3062-3069.  DOI: 10.11772/j.issn.1001-9081.2022091449

Asbtract ( )   HTML ( )   PDF (1411KB) ( )  

Figures and Tables | References | Related Articles | Metrics

As one of the core technologies of natural language processing， sentence embedding affects the quality and performance of natural language processing system. However， the existing methods are unable to infer the global semantic relationship between sentences efficiently， which leads to the fact that the semantic similarity measurement of sentences in Euclidean space still has some problems. To address the issue， a sentence embedding optimization method based on manifold learning was proposed. In the method， Local Linear Embedding （LLE） was used to perform double weighted local linear combinations to the sentences and their semantically similar sentences， thereby preserving the local geometric information between sentences and providing helps to the inference of the global geometric information. As a result， the semantic similarity of sentences in Euclidean space was closer to the real semantics of humans. Experimental results on seven text semantic similarity tasks show that the proposed method has the average Spearman’s Rank Correlation Coefficient， （SRCC） improved by 1.21 percentage points compared with the contrastive learning-based method SimCSE （Simple Contrastive learning of Sentence Embeddings）. In addition， the proposed method was applied to mainstream pre-trained models. The results show that compared to the original pre-trained models， the models optimized by the proposed method have the average SRCC improved by 3.32 to 7.70 percentage points.

Text semantic de-duplication algorithm based on keyword graph representation

Jinyun WANG, Yang XIANG

2023, 43(10):  3070-3076.  DOI: 10.11772/j.issn.1001-9081.2022101495

Asbtract ( )   HTML ( )   PDF (1266KB) ( )  

Figures and Tables | References | Related Articles | Metrics

There are a large number of redundant texts with the same or similar semantics in the network. Text de-duplication can solve the problem that redundant texts waste storage space and can reduce unnecessary consumption for information extraction tasks. Traditional text de-duplication algorithms rely on literal overlapping information， and do not make use of the semantic information of texts； at the same time， they cannot capture the interaction information between sentences that are far away from each other in long text， so that the de-duplication effect of these methods is not ideal. Aiming at the problem of text semantic de-duplication， a long text de-duplication algorithm based on keyword graph representation was proposed. Firstly， the text pair was represented as a graph with the keyword phrase as the vertex by extracting the semantic keyword phrase from the text pair. Secondly， the nodes were encoded in various ways， and Graph Attention Network （GAT） was used to learn the relationship between nodes to obtain the vector representation of text to the graph， and judge whether the text pairs were semantically similar. Finally， the de-duplication processing was performed according to the text pair’s semantical similarity. Compared with the traditional methods， this method can use the semantic information of texts effectively， and through the graph structure， the method can connect the distant sentences in the long text by the co-occurrence relationship of keyword phrases to increase the semantic interaction between different sentences. Experimental results show that the proposed algorithm performs better than the traditional algorithms， such as Simhash， BERT （Bidirectional Encoder Representations from Transformers） fine-tuning and Concept Interaction Graph （CIG）， on both CNSE （Chinese News Same Event） and CNSS （Chinese News Same Story） datasets. Specifically， the F1 score of the proposed algorithm on CNSE dataset is 84.65%， and that on CNSS dataset reaches 90.76%. The above indicates that the proposed algorithm can improve the effect of text de-duplication tasks effectively.

High-low dimensional feature guided real-time semantic segmentation network

Zixing YU, Shaojun QU, Xin HE, Zhuo WANG

2023, 43(10):  3077-3085.  DOI: 10.11772/j.issn.1001-9081.2022091438

Asbtract ( )   HTML ( )   PDF (1967KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Most semantic segmentation networks use bilinear interpolation to restore the resolution of the high-level feature map to the same resolution as the low-level feature map and then perform fusion operation， which causes that part of high-level semantic information cannot be spatially aligned with the low-level feature map， resulting in the loss of semantic information. To solve the problem， based on the improvement of Bilateral Segmentation Network （BiSeNet）， a High-Low dimensional Feature Guided real-time semantic segmentation Network （HLFGNet） was proposed. First， High-Low dimensional Feature Guided Module （HLFGM） was proposed to guide the displacement of high-level semantic information during the upsampling process through the spatial position information of the low-level feature map. At the same time， the strong feature representations were obtained by the high-level feature maps， and by combining with the attention mechanism， the redundant edge detail information in the low-level feature map was eliminated and the pixel misclassification was reduced. Then， the improved Pyramid Pooling Guided Module （PPGM） was introduced to obtain global contextual information and strengthen the effective fusion of local contextual information at different scales. Experimental results on Cityscapes validation set and CamVid test set show that HLFGNet has the mean Intersection over Union （mIoU） of 76.67% and 70.90% respectively， the frames per second reached 75.0 and 96.2 respectively. In comparison with BiSeNet， HLFGNet has the mIoU increased by 1.76 and 3.40 percentage points respectively. It can be seen that HLFGNet can accurately identify the scene information and meet the real-time requirements.

Aspect sentiment analysis with aspect item and context representation

Dan XU, Hongfang GONG, Rongrong LUO

2023, 43(10):  3086-3092.  DOI: 10.11772/j.issn.1001-9081.2022101482

Asbtract ( )   HTML ( )   PDF (1011KB) ( )  

Figures and Tables | References | Related Articles | Metrics

When predicting the emotional polarity of a specific aspect， there is a problem of only depending on a single aspect item and ignoring the emotional dependence between aspect items in the same sentence， a Multi-layer Multi-hop Memory network with Aspect Item and Context Representation （AICR-M³net） was proposed. Firstly， the position weighting information was fused by Bi-directional Gated Recurrent Unit （Bi-GRU）， and the hidden layer output was used as the input of the mixed context coding layer to obtain a context representation with higher semantic relevance to the context. Then， Multi-layer Multi-hop Memory Networks （M³net） was introduced to match aspect words and context many times and word by word to generate aspect word vectors of specific context. At the same time， the emotional dependence between specific aspect item and other aspect items in the sentence was modeled to guide the generation of context vector of specific aspect item. Experimental results on Restaurant， Laptop and Twitter datasets show that the proposed model has the classification accuracy improved by 1.34， 3.05 and 2.02 percentage points respectively， and the F1 score increased by 3.90， 3.78 and 2.94 percentage points respectively， compared with AOA-MultiACIA （Attention-Over-Attention Multi-layer Aspect-Context Interactive Attention）. The above verifies that the proposed model can deal with the mixed information with multiple aspects in context more effectively， and has certain advantages in dealing with the sentiment classification task in specific aspects.

Text adversarial example generation method based on BERT model

Yuhang LI, Yuli YANG, Yao MA, Dan YU, Yongle CHEN

2023, 43(10):  3093-3098.  DOI: 10.11772/j.issn.1001-9081.2022091468

Asbtract ( )   HTML ( )   PDF (971KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that the existing adversarial example generation methods require a lot of queries to the target model， which leads to poor attack effects， a Text Adversarial Examples Generation Method based on BERT （Bidirectional Encoder Representations from Transformers） model （TAEGM） was proposed. Firstly， the attention mechanism was adopted to locate the keywords that significantly influence the classification results without query of the target model. Secondly， word-level perturbation of keywords was performed by BERT model to generate candidate adversarial examples. Finally， the candidate examples were clustered， and the adversarial examples were selected from the clusters that have more influence on the classification results. Experimental results on Yelp Reviews， AG News， and IMDB Review datasets show that compared to the suboptimal adversarial example generation method CLARE （ContextuaLized AdversaRial Example generation model） on Success Rate （SR）， TAEGM can reduce the Query Counts （QC） to the target model by 62.3% and time consumption by 68.6% averagely while ensuring the SR of adversarial attacks. Based on the above， further experimental results verify that the adversarial examples generated by TAEGM not only have good transferability， but also improve the robustness of the model through adversarial training.

Dynamic evaluation method for benefit of modality augmentation

Yizhen BI, Huan MA, Changqing ZHANG

2023, 43(10):  3099-3106.  DOI: 10.11772/j.issn.1001-9081.2022101510

Asbtract ( )   HTML ( )   PDF (2673KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Focused on the difficulty and big benefit difference in acquiring new modalities， a method for dynamically evaluating benefit of modality augmentation was proposed. Firstly， the intermediate feature representation and the prediction results before and after modality fusion were obtained through the multimodal fusion network. Then， the confidence before and after fusion were obtained by introducing the True Class Probability （TCP） of two prediction results to confidence estimation. Finally， the difference between two confidences was calculated and used as an sample to obtain the benefit brought by the new modality. Extensive experiments were conducted on commonly used multimodal datasets and real medical datasets such as The Cancer Genome Atlas （TCGA）. The experimental results on TCGA dataset show that compared with the random benefit evaluation method and the Maximum Class Probability （MCP） based method， the proposed method has the accuracy increased by 1.73 to 4.93 and 0.43 to 4.76 percentage points respectively， and the Effective Sample Rate （ESR） increased by 2.72 to 11.26 and 1.08 to 25.97 percentage points respectively. It can be seen that the proposed method can effectively evaluate benefits of acquiring new modalities for different samples， and has a certain degree of interpretability.

TenrepNN：practice of new ensemble learning paradigm in enterprise self-discipline evaluation

Jingtao ZHAO, Zefang ZHAO, Zhaojuan YUE, Jun LI

2023, 43(10):  3107-3113.  DOI: 10.11772/j.issn.1001-9081.2022091454

Asbtract ( )   HTML ( )   PDF (1741KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In order to cope with the current situations of low self-discipline， frequent violation events and difficult government supervision of enterprises in the internet environment， a Two-layer ensemble residual prediction Neural Network （TenrepNN） model was proposed to evaluate the self-discipline of enterprises. And by integrating the ideas of Stacking and Bagging ensemble learning， a new paradigm of integrated learning was designed， namely Adjusting. TenrepNN model has a two-layer structure. In the first layer， three base learners were used to predict the enterprise score preliminarily. In the second layer， the idea of residual correction was adopted， and a residual prediction neural network was proposed to predict the output deviation of each base learner. Finally， the final output was obtained by adding the deviations and the base learner scores together. On the enterprise self-discipline evaluation dataset， compared with the traditional neural network， the proposed model has the Root Mean Square Error （RMSE） reduced by 2.7%， and the classification accuracy in the self-discipline level reached 94.51%. Experimental results show that by integrating different base learners to reduce the variance and using residual prediction neural network to decrease the deviation explicitly， TenrepNN model can accurately evaluate enterprise self-discipline to achieve differentiated dynamic supervision.

Highway traffic flow prediction based on feature fusion graph attention network

Chun GAO, Mengling WANG

2023, 43(10):  3114-3120.  DOI: 10.11772/j.issn.1001-9081.2022101587

Asbtract ( )   HTML ( )   PDF (2825KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Based on the actual spatio-temporal topology of the traffic network， a Feature Fusion Graph ATtention network （FF-GAT） model was proposed to fuse multiple traffic state information obtained by nodes， so as to predict the highway traffic flow. First， the correlation features among the vehicle speed， traffic flow and occupancy of the nodes were analyzed， and based on the multivariate temporal attention mechanism， the relationships among the vehicle speed， traffic flow and occupancy were incorporated into the attention mechanism to capture the dynamic temporal correlation between different moments of traffic flow. Then， the nodes were divided into different sets of neighborhoods， and the spatial correlation between different neighborhoods of traffic flow was captured by the feature fusion Graph Attention neTwork （GAT）. At the same time， the coupling correlation between multiple heterogeneous data was fully explored by the feature crossover network to provide effective information supplement for predicting the target sequence. Experiments were carried out on two publicly available traffic flow datasets. Experimental results show that FF-GAT model reduces the Root Mean Squared Error （RMSE） by 3.4% compared with ASTGCN （Attention based Spatial-Temporal Graph Convolutional Network） model and 3.1% compared with GCN-GAN （Graph Convolutional Network and Generative Adversarial Network） model on PeMSD8 dataset. It can be seen that FF-GAT model can effectively improve the prediction accuracy through feature fusion.

Data science and technology

Fuzzy-rough set based unsupervised dynamic feature selection algorithm

Lei MA, Chuan LUO, Tianrui LI, Hongmei CHEN

2023, 43(10):  3121-3128.  DOI: 10.11772/j.issn.1001-9081.2022101543

Asbtract ( )   HTML ( )   PDF (511KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Dynamic feature selection algorithms can improve the time efficiency of processing dynamic data. Aiming at the problem that there are few unsupervised dynamic feature selection algorithms based on fuzzy-rough sets， an Unsupervised Dynamic Fuzzy-Rough set based Feature Selection （UDFRFS） algorithm was proposed under the condition of features arriving in batches. First， by defining a pseudo triangular norm and new similarity relationship， the process of updating fuzzy relation value was performed on the basis of existing data to reduce unnecessary calculation. Then， by utilizing the existing feature selection results， dependencies were adopted to judge if the original feature part would be recalculated to reduce the redundant process of feature selection， and the feature selection was further speeded up. Experimental results show that compared to the static dependency-based unsupervised fuzzy-rough set feature selection algorithm， UDFRFS can achieve the time efficiency improvement of more than 90 percentage points with good classification accuracy and clustering performance.

Spectral clustering based dynamic community discovery algorithm in social network

Yu YANG, Weiwei DUAN

2023, 43(10):  3129-3135.  DOI: 10.11772/j.issn.1001-9081.2022101517

Asbtract ( )   HTML ( )   PDF (2785KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Dynamic community discovery is an important research area in Social Network Analysis （SNA）. As nodes joining or leaving social networks， the relationships between nodes establish or terminate， which affects community structure changes. The discovery algorithms of static communities in social networks lack of the essential historical information of community nodes， resulting in the insufficient network structure analysis as well as clustering information and the high computational cost. Aiming at these problems， based on the division of the community network evolution events， according to the analysis of the major community events， a Spectral Clustering based Dynamic Community Discovery Algorithm （SC-DCDA） was proposed. Firstly， according to the experimental observation， the dimensionality of high-dimensional data was reduced by using the method of spectral mapping. At the same time， the improved Fuzzy C-Means clustering （FCM） algorithm was adopted to determine the correlation between the nodes in the dynamic social network and the communities to be discovered. Secondly， the community structures were analyzed according to the evolutionary similarity matrix. Finally， the real network datasets and community discovery algorithm indicators， such as modularity score and Silhouette coefficient， were used to evaluate the effects of the proposed algorithm. Experimental results show that the computational cost of SC-DCDA is reduced by 8.37% compared with traditional spectral clustering， the average modularity score of the algorithm on all datasets is 0.49， and the qualitative analysis results of other algorithm metrics are also good， indicating that the proposed algorithm performs well in information interaction， clustering effect， and accuracy.

Collaborative filtering algorithm based on collaborative training and Boosting

Xiaohan YANG, Guosheng HAO, Xiehua ZHANG, Zihao YANG

2023, 43(10):  3136-3141.  DOI: 10.11772/j.issn.1001-9081.2022101489

Asbtract ( )   HTML ( )   PDF (1305KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Collaborative Filtering （CF） algorithm can realize personalized recommendation on the basis of the similarity between items or users. However， data sparsity has always been one of the challenges faced by CF algorithm. In order to improve the prediction accuracy， a CF algorithm based on Collaborative Training and Boosting （CFCTB） was proposed to solve the problem of sparse user-item scores. First， two CFs were integrated into a framework by using collaborative training， pseudo-labeled samples with high confidence were added to each other’s training set by the two CFs， and Boosting weighted training data were used to assist the collaborative training. Then， the weighted integration was used to predict the final user scores， and the accumulation of noise generated by pseudo-labeled samples was avoided effectively， thereby further improving the recommendation performance. Experimental results show that the accuracy of the proposed algorithm is better than that of the single models on four open datasets. On CiaoDVD dataset with the highest sparsity， compared with Global and Local Kernels for recommender systems （GLocal-K）， the proposed algorithm has the Mean Absolute Error （MAE） reduced by 4.737%. Compared with ECoRec （Ensemble of Co-trained Recommenders） algorithm， the proposed algorithm has the Root Mean Squared Error （RMSE） decreased by 7.421%. The above rasults verify the effectiveness of the proposed algorithm.

Cyber security

Low-cost pay-per-use licensing scheme for FPGA intellectual property protection

Binwei SONG, Yao WANG

2023, 43(10):  3142-3148.  DOI: 10.11772/j.issn.1001-9081.2022101506

Asbtract ( )   HTML ( )   PDF (1413KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The pay-per-use licensing of the Intellectual Property （IP） core enables the system designer to purchase IP at low price according to the actual situation， and has become a major method of IP licensing. To meet the pay-per-use demand of IP core， based on Reconfigurable Finite State Machine （RFSM） and Physical Unclonable Function （PUF）， a new IP licensing scheme RFSM-PUF was proposed for Field Programmable Gate Array （FPGA） IP. Aiming at the problem that the protocols of the IP protection schemes of different manufacturers cannot be used universally， an IP protection authentication protocol for the proposed scheme was proposed to ensure the confidentiality and flexibility of IP authentication. Firstly， RFSM was embedded in the Original Finite State Machine （OFSM） in the IP， and in this way， the IP was only unlocked by the IP core designer. Then， the challenges were input into the PUF circuit to produce responses. Finally， the cipher consisting of the license and PUF responses was input into the RFSM to unlock the IP. The security analysis results show that the proposed scheme meets various security indicators. RFSM-PUF scheme was tested on the LGSyth91 benchmark circuits. Experimental results show that on the premise of meeting various safety indicators， the proposed scheme reduces 1 377 Look-Up Tables （LUT） averagely at every IP core compared to the PUF based pay-per-use licensing scheme， so that the hardware overhead is significantly reduced.

Medical image privacy protection based on thumbnail encryption and distributed storage

Na ZHOU, Ming CHENG, Menglin JIA, Yang YANG

2023, 43(10):  3149-3155.  DOI: 10.11772/j.issn.1001-9081.2022111646

Asbtract ( )   HTML ( )   PDF (4138KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the popularity of cloud storage services and telemedicine platforms， more and more medical images are uploaded to the cloud. After being uploaded， the uploaded medical images may be leaked to unauthorized third parties， resulting in the disclosure of users’ personal privacy. Besides， if medical images are only uploaded to a single server for storage， they are vulnerable to attacks resulting in the loss of all data. To solve these problems， a medical image privacy protection algorithm based on thumbnail encryption and distributed storage was proposed. Firstly， by encrypting the thumbnail of the original medical image， the relevance of the medical images was preserved properly while achieving the encryption effect. Secondly， the double embedding method was adopted when hiding secret information， and data extraction and image recovery were performed separately to achieve Reversible Data Hiding （RDH） of the encrypted image. Finally， the distributed storage method based on polynomial shared matrix was used to generate n shares of the image and distribute them to n servers. Experimental results show that by using the encrypted thumbnail as carrier， the proposed algorithm exceeds the traditional security encryption methods on embedding rate. Even if the server is attacked， the receiver can recover the original image and private information as long as it receives no less than k shares. In the privacy protection of medical images， experiments were carried out from the aspects of anti-attack and image recovery， and the analysis results show that the proposed encryption algorithm has good performance and high security.

Design and implementation of cipher component security criteria testing tool

Shanshan HUO, Yanjun LI, Jian LIU, Yinshuang LI

2023, 43(10):  3156-3161.  DOI: 10.11772/j.issn.1001-9081.2022091443

Asbtract ( )   HTML ( )   PDF (2718KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Symmetric cryptography is the core technology of data confidentiality in information systems. At the same time， nonlinear S-box is usually the key cryptographic component， and is widely used in the design of block cipher， stream cipher， MAC （Message Authentication Code） algorithm， etc. In order to ensure the security of the cryptographic algorithm design， firstly， the criteria testing methods for differential uniformity， nonlinearity， fixed point number， algebraic degree and item number， algebraic immunity， avalanche characteristic and diffusion characteristic were researched. Secondly， the results of each security criterion of the S-box were designed and output in the visual window， and the detailed descriptions of the corresponding security criterion were given in a pop-up window way. Thirdly， the design of the sub-components of nonlinearity and algebraic immunity was focused， and the linear distribution table was simplified according to the nonlinearity. At the same time， based on the theorem， the calculation process of algebraic immunity was optimized and illustrated with an example. Finally， the S-box testing tool was implemented with seven security criteria， and the test cases were demonstrated. The proposed tool is mainly used to test the security criteria of the nonlinear component S-box in the symmetric cryptographic algorithm， and then provides a guarantee for the security of the overall algorithm.

Advanced computing

Survey of data-driven intelligent cloud-edge collaboration

Pengxin TIAN, Guannan SI, Zhaoliang AN, Jianxin LI, Fengyu ZHOU

2023, 43(10):  3162-3169.  DOI: 10.11772/j.issn.1001-9081.2022091418

Asbtract ( )   HTML ( )   PDF (1772KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the rapid development of Internet of Things （IoT）， a large amount of data generated in edge scenarios such as sensors often needs to be transmitted to cloud nodes for processing， which brings huge transmission cost and processing delay. Cloud-edge collaboration provides a solution for these problems. Firstly， on the basis of comprehensive investigation and analysis of the development process of cloud-edge collaboration， combined with the current research ideas and progress of intelligent cloud-edge collaboration， the data acquisition and analysis， computation offloading technology and model-based intelligent optimization technology in cloud edge architecture were analyzed and discussed emphatically. Secondly， the functions and applications of various technologies in intelligent cloud-edge collaboration were analyzed deeply from the edge and the cloud respectively， and the application scenarios of intelligent cloud-edge collaboration technology in reality were discussed. Finally， the current challenges and future development directions of intelligent cloud-edge collaboration were pointed out.

Compilation optimizations for inconsistent control flow on deep computer unit

Xiaoyi YANG, Rongcai ZHAO, Hongsheng WANG, Lin HAN, Kunkun XU

2023, 43(10):  3170-3177.  DOI: 10.11772/j.issn.1001-9081.2022091338

Asbtract ( )   HTML ( )   PDF (4315KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The domestic DCU （Deep Computer Unit） adopts the parallel execution model of Single Instruction Multiple Thread （SIMT）. When the programs are executed， inconsistent control flow is generated in the kernel function， which causes the threads in the warp be executed serially. And that is warp divergence. Aiming at the problem that the performance of the kernel function is severely restricted by warp divergence， a compilation optimization method to reduce the warp divergence time — Partial-Control-Flow-Merging （PCFM） was proposed. Firstly， divergence analysis was performed to find the fusible divergent regions that are isomorphic and contained a large number of same instructions and similar instructions. Then， the fusion profit of the fusible divergent regions was evaluated by counting the percentage of instruction cycles saved after merging. Finally， the alignment sequence was searched， the profitable fusible divergent regions were merged. Some test cases from Graphics Processing Unit （GPU） benchmark suite Rodinia and the classic sorting algorithm were selected to test PCFM on DCU. Experimental results show that PCFM can achieve an average speedup ratio of 1.146 for the test cases. And the speedup of PCFM is increased by 5.72% compared to that of the branch fusion + tail merging method. It can be seen that the proposed method has a better effect on reducing warp divergence.

Dynamic multi-objective optimization algorithm based on adaptive prediction of new evaluation index

Erchao LI, Shenghui ZHANG

2023, 43(10):  3178-3187.  DOI: 10.11772/j.issn.1001-9081.2022091453

Asbtract ( )   HTML ( )   PDF (3391KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Most of the Multi-objective Optimization Problems （MOP） in real life are Dynamic Multi-objective Optimization Problems （DMOP）， and the objective function， constraint conditions and decision variables of such problems may change with time， which requires the algorithm to quickly adapt to the new environment after the environment changes， and guarantee the diversity of Pareto solution sets while converging to the new Pareto frontier quickly. To solve the problem， an Adaptive Prediction Dynamic Multi-objective Optimization Algorithm based on New Evaluation Index （NEI-APDMOA） was proposed. Firstly， a new evaluation index better than crowding was proposed in the process of population non-dominated sorting， and the convergence speed and population diversity were balanced in different stages， so as to make the convergence process of population more reasonable. Secondly， a factor that can judge the strength of environmental changes was proposed， thereby providing valuable information for the prediction stage and guiding the population to better adapt to environmental changes. Finally， three more reasonable prediction strategies were matched according to environmental change factor， so that the population was able to respond to environmental changes quickly. NEI-APDMOA， DNSGA-Ⅱ-A （Dynamic Non-dominated Sorting Genetic Algorithm-Ⅱ-A）， DNSGA-Ⅱ-B （Dynamic Non-dominated Sorting Genetic Algorithm-Ⅱ-B） and PPS （Population Prediction Strategy） algorithms were compared on nine standard dynamic test functions. Experimental results show that NEI-APDMOA achieves the best average Inverted Generational Distance （IGD） value， average SPacing （SP） value and average Generational Distance （GD） value on nine， four and eight test functions respectively， and can respond to environmental changes faster.

Network and communications

Trade-off between energy efficiency and spectrum efficiency for decode-and-forward full-duplex relay network

Qian ZHANG, Runhe QIU

2023, 43(10):  3188-3194.  DOI: 10.11772/j.issn.1001-9081.2022091414

Asbtract ( )   HTML ( )   PDF (1778KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In order to optimize the Energy Efficiency （EE） and Spectrum Efficiency （SE） of Decode-and-Forward （DF） full-duplex relay network， a trade-off method of EE and SE for DF full-duplex relay network was proposed. In full-duplex relay network， firstly， the EE of the network was optimized with the goal of improving the SE of the network. And the optimal power of the relay was obtained by combining the derivation and the Newton-Raphson method， then the Pareto optimal set of the objective function was given. Secondly， a trade-off factor was introduced through the weighted scalar method， a trade-off optimization function of EE and SE was constructed， and the multi-objective optimization problem of EE optimization and SE optimization was transformed into a single-objective energy-spectrum efficiency optimization problem by using normalization. At the same time， the performance of EE， SE and trade-off optimization under different trade-off factor was analyzed. Simulation results show that the SE and EE of the proposed method are higher at the same data transmission rate compared with the those of the full-duplex-optimal power method and the half-duplex-optimal relay-optimal power allocation method. By adjusting different trade-off factors， the optimal trade-off and the optimization of EE and SE can be achieved.

Optimized bit allocation algorithm for coding tree unit level

Xu YANG, Hongwei GUO, Wanxue LI

2023, 43(10):  3195-3201.  DOI: 10.11772/j.issn.1001-9081.2022091410

Asbtract ( )   HTML ( )   PDF (742KB) ( )  

Figures and Tables | References | Related Articles | Metrics

It the rate control algorithms of the new generation video coding standard H.266/VVC （Versatile Video Coding）， the rate-distortion optimization technique with independent coding parameters is adopted. However， the Coding Tree Units （CTUs） within the same frame affect others in the spatial domain， and there are global coding parameters. At the same time， in the CTU-level bit allocation formulas， approximated coding parameters for bit allocation are used， resulting in the reduction of rate control accuracy and coding performance. To address these issues， a spatial-domain global optimization algorithm for CTU-level bit allocation called RTE_RC （Rate Control with Recursive Taylor Expansion） was proposed， and the global coding parameters were approximated by using a recursive algorithm. Firstly， a globally optimized bit allocation model in spatial-domain was established. Secondly， a recursive algorithm was used to calculate the global Lagrange multiplier in the CTU-level bit allocation formula. Finally， the bit allocation of coding units was optimized and the coding units were coded. Experimental results show that under the Low-Delay Prediction frame （LDP） configuration， compared with the rate control algorithm VTM_RC （Rate Control algorithm Versatile Test Model）， the proposed algorithm has the rate control error decreased from 0.46% to 0.02%， the bit-rate saved by 2.48 percentage points， and the coding time reduced by 3.52%. Therefore， the rate control accuracy and rate distortion performance are significantly improved by the proposed algorithm.

Multimedia computing and computer simulation

Semantic segmentation of point cloud scenes based on multi-feature fusion

Wen HAO, Yang WANG, Hainan WEI

2023, 43(10):  3202-3208.  DOI: 10.11772/j.issn.1001-9081.2023020119

Asbtract ( )   HTML ( )   PDF (2678KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In order to mine the semantic relationships and spatial distribution among features， and further improve the semantic segmentation results of point cloud through multi-feature enhancement， a Multi-Feature Fusion based point cloud scene semantic segmentation Network （MFF-Net） was proposed. In the proposed network， the 3D coordinates and improved edge features were used as input， firstly， the neighbor points of the point were searched by using K-Nearest Neighbor （KNN） algorithm， and the geometric offsets were calculated based on 3D coordinates and coordinate differences among neighbor points， which enhanced the local geometric feature representation of points. Secondly， the distance between the central point and its neighbor points were used to as weighting information to update the edge features， and the spatial attention mechanism was introduced to obtain the semantic information among features. Thirdly， the spatial distribution information among features was further extracted by calculating the differences among neighbor features and using mean pooling operation. Finally， the trilateral features were fused based on attention pooling. Experimental results demonstrate that on S3DIS （Stanford 3D large-scale Indoor Spaces） dataset， the mean Intersection over Union （mIoU） of the proposed network is 67.5%， and the Overall Accuracy （OA） of the proposed network is 87.2%. These two values are 10.2 and 3.4 percentage points higher than those of PointNet++ respectively. It can be seen that MFF-Net can achieve good segmentation results in both large indoor and outdoor scenes.

Multi-contour segmentation algorithm for point cloud slices of irregular objects

Jin ZHANG, Wen XU, Yuqiao ZHOU, Kai LIU

2023, 43(10):  3209-3216.  DOI: 10.11772/j.issn.1001-9081.2022101536

Asbtract ( )   HTML ( )   PDF (4343KB) ( )  

Figures and Tables | References | Related Articles | Metrics

When using the slicing method to measure the point cloud volumes of irregular objects， the existing Polygon Splitting and Recombination （PSR） algorithm cannot split the nearer contours correctly， resulting in low calculation precision. Aiming at this problem， a multi-contour segmentation algorithm — Improved Nearest Point Search （INPS） algorithm was proposed. Firstly， the segmentation of multiple contours was performed through the single-use principle of local points. Then， Point Inclusion in Polygon （PIP） algorithm was adopted to judge the inclusion relationship of contours， thereby determining positive or negative property of the contour area. Finally， the slice area was multiplied by the thickness and the results were accumulated to obtain the volume of irregular object point cloud. Experimental results show that on two public point cloud datasets and one point cloud dataset of chemical electron density isosurface， the proposed algorithm can achieve high-accuracy boundary segmentation and has certain universality. The average relative error of volume measurement of the proposed algorithm is 0.043 6%， which is lower than 0.062 7% of PSR algorithm， verifying that the proposed algorithm achieves high accuracy boundary segmentation.

Double complex convolution and attention aggregating recurrent network for speech enhancement

Bennian YU, Yongzhao ZHAN, Qirong MAO, Wenlong DONG, Honglin LIU

2023, 43(10):  3217-3224.  DOI: 10.11772/j.issn.1001-9081.2022101533

Asbtract ( )   HTML ( )   PDF (1993KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems of limited representation of spectrogram feature correlation information and unsatisfactory denoising effect in the existing speech enhancement methods， a speech enhancement method of Double Complex Convolution and Attention Aggregating Recurrent Network （DCCARN） was proposed. Firstly， a double complex convolutional network was established to encode the two-branch information of the spectrogram features after the short-time Fourier transform. Secondly， the codes in the two branches were used in the inter- and and intra-feature-block attention mechanisms respectively， and different speech feature information was re-labeled. Secondly， the long-term sequence information was processed by Long Short-Term Memory （LSTM） network， and the spectrogram features were restored and aggregated by two decoders. Finally， the target speech waveform was generated by short-time inverse Fourier transform to achieve the purpose of suppressing noise. Experiments were carried out on the public dataset VBD （Voice Bank+DMAND） and the noise added dataset TIMIT. The results show that compared with the phase-aware Deep Complex Convolution Recurrent Network （DCCRN）， DCCARN has the Perceptual Evaluation of Speech Quality （PESQ） increased by 0.150 and 0.077 to 0.087 respectively. It is verified that the proposed method can capture the correlation information of spectrogram features more accurately， suppress noise more effectively， and improve speech intelligibility.

Robot hand-eye calibration algorithm based on covariance matrix adaptation evolutionary strategy

Yuntao ZHAO, Wanqi XIE, Weigang LI, Jiaming HU

2023, 43(10):  3225-3229.  DOI: 10.11772/j.issn.1001-9081.2022081282

Asbtract ( )   HTML ( )   PDF (1281KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To solve the problem that the traditional hand-eye calibration algorithms have large solution errors due to the noise interference in the processes of vision sensor calibration and robot kinematics solution， a robot hand-eye calibration algorithm based on Covariance Matrix Adaptation Evolutionary Strategy （CMAES） was proposed. Firstly， the mathematical tool Dual Quaternion （DQ） was used to establish the objective functions and geometric constraints for both rotation and translation， and the solution model was simplified. Then， the penalty function method was used to transform the constrained problem into an unconstrained optimization problem. Finally， CMAES algorithm was used to approximate the global optimal solution of hand-eye calibration rotation and translation equations. An experimental platform of robot and camera measurement was built， and the proposed algorithm was compared with two-step Tsai algorithm， the nonlinear optimization algorithm INRIA， and the DQ algorithm. Experimental results show that the solution error and variance of the proposed algorithm are smaller than those of traditional algorithms for both rotation and translation. Compared with Tsai algorithm， the proposed algorithm has the rotation accuracy improved by 4.58%， and the translation accuracy improved by 10.54%. It can be seen that the proposed algorithm has better solution accuracy and stability in the actual hand-eye calibration process with noise interference.

Stereo matching algorithm based on cross-domain adaptation

Chuanbiao LI, Yuanwei BI

2023, 43(10):  3230-3235.  DOI: 10.11772/j.issn.1001-9081.2022091398

Asbtract ( )   HTML ( )   PDF (2684KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Convolutional Neural Networks （CNNs） have made good progress in supervised stereo matching tasks， but most CNN algorithms are difficult to perform well in cross-domain situations. Aiming at the stereo matching problem of cross-domain data， a Cross-domain Adaptation Stereo Matching Network （CASM-Net） algorithm was proposed to achieve domain adaptive stereo matching tasks using transfer learning based on CNN. In the algorithm， a transferable feature extraction module was used to extract rich wide-domain features for stereo matching tasks. At the same time， an adaptive cost optimization module was designed to obtain the optimal cost distribution by making use of the similarity information on different receptive fields to optimize the cost. In addition， a disparity score prediction module was proposed to quantify the stereo matching ability of different regions， and the disparity results were further optimized by adjusting the disparity search range of the image. Experimental results show that on KITTI2012 and KITTI2015 datasets， compared with PSMNet （Pyramid Stereo Matching Network） algorithm， CASM-Net algorithm reduces 6.1%， 3.3% and 19.3% in 2-PE-Noc， 2-PE-All and 3-PE-fg， respectively； on Middlebury dataset， without re-training， CASM-Net algorithm achieves the optimal or suboptimal 2-PE results on all samples in the comparison with other algorithms. It can be seen that CASM-Net algorithm can improve cross-domain stereo matching.

Human action recognition method based on multi-scale feature fusion of single mode

Suolan LIU, Zhenzhen TIAN, Hongyuan WANG, Long LIN, Yan WANG

2023, 43(10):  3236-3243.  DOI: 10.11772/j.issn.1001-9081.2022101473

Asbtract ( )   HTML ( )   PDF (1425KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In order to solve the problem of insufficient mining of potential association between remote nodes in human action recognition tasks， and the problem of high training cost caused by using multi-modal data， a multi-scale feature fusion human action recognition method under the condition of single mode was proposed. Firstly， the global feature correlation of the original skeleton diagram of human body was carried out， and the coarse-scale global features were used to capture the connections between the remote nodes. Secondly， the global feature correlation graph was divided locally to obtain the Complementary Subgraphs with Global Features （CSGFs）， the fine-scale features were used to establish the strong correlation， and the multi-scale feature complementarity was formed. Finally， the CSGFs were input into the spatial-temporal Graph Convolutional module for feature extraction， and the extracted results were aggregated to output the final classification results. Experimental results show that the accuracy of the proposed method on the authoritative action recognition dataset NTU RGB+D60 is 89.0% （X-sub） and 94.2% （X-view） respectively. On the challenging large-scale dataset NTU RGB+D120， the accuracy of the proposed method is 83.3% （X-sub） and 85.0% （X-setup） respectively， which is 1.4 and 0.9 percentage points higher than that of the ST-TR （Spatial-Temporal TRansformer） under single modal respectively， and 4.1 and 3.5 percentage points higher than that of the lightweight SGN （Semantics-Guided Network）. It can be seen that the proposed method can fully exploit the synergistic complementarity of multi-scale features， and effectively improve the recognition accuracy and training efficiency of the model under the condition of single modal.

Moving portrait debluring network based on multi-level jump residual group

Jiaqi JI, Zhenkun LU, Fupeng XIONG, Tian ZHANG, Hao YANG

2023, 43(10):  3244-3250.  DOI: 10.11772/j.issn.1001-9081.2022091457

Asbtract ( )   HTML ( )   PDF (3316KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To address the issues of blurred contours and lost details of portrait image with motion blur after restoration， a moving portrait deblurring method based on multi-level jump residual group Generation Adversarial Network （GAN） was proposed. Firstly， the residual block was improved to construct the multi-level jump residual group module， and the structure of PatchGAN was also improved to make GAN better combine with the image features of each layer. Secondly， the multi-loss fusion method was adopted to optimize the network to enhance the real texture of the reconstructed image. Finally， the end-to-end mode was used to perform blind deblurring on the motion blurred portrait image and output clear portrait image. Experimental results on CelebA dataset show that the Peak Signal-to-Noise Ratio （PSNR） and Structural SIMilarity （SSIM） of the proposed method are at least 0.46 dB and 0.05 higher than those of the Convolutional Neural Network （CNN）-based methods such as DeblurGAN （Deblur GAN）， Scale-Recurrent Network （SRN） and MSRAN （Multi-Scale Recurrent Attention Network）. At the same time， the proposed method has fewer model parameters， faster restoration， and more texture details in the restored portrait images.

Dual U-Former image deraining network based on non-separable lifting wavelet

Bin LIU, Siyan FANG

2023, 43(10):  3251-3259.  DOI: 10.11772/j.issn.1001-9081.2022091422

Asbtract ( )   HTML ( )   PDF (5959KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that the deraining methods based on tensor product wavelet cannot capture high-frequency rain streaks in all directions， a Dual U-Former Network （DUFN） based on non-separable lifting wavelet was proposed. Firstly， the isotropic non-separable lifting wavelet was used to capture high-frequency rain streaks in all directions. In this way， compared with tensor product wavelets such as Haar wavelet， which can only capture high-frequency rain streaks in three directions， DUFN was able to obtain more comprehensive rain streak information. Secondly， two U-Nets composed of Transformer Blocks （TBs） were connected in series at various scales， so that the semantic features of the shallow decoder were transferred to the deep stage， and the rain streaks were removed more thoroughly. At the same time， the scale-guide encoder was used to guide the coding stage by using the information of various scales in the shallow layer， and Gated Fusion Module （GFM） based on CBAM （Convolutional Block Attention Module） was used to make the fusion process put more focus on the rain area. Experimental results on Rain200H， Rain200L， Rain1200 and Rain12 synthetic datasets show that the Structure SIMilarity （SSIM） of DUFN is improved by 0.009 7 on average compared to that of the advanced method SPDNet （Structure-Preserving Deraining Network）. And on Rain200H， Rain200L and Rain12 synthetic datasets， the Peak Signal-to-Noise Ratio （PSNR） of DUFN is improved by 0.657 dB averagely. On real-world dataset SPA-Data， PSNR and SSIM of DUFN are improved by 0.976 dB and 0.003 1 respectively compared with those of the advanced method ECNetLL （Embedding Consistency Network+Layered Long short-term memory）. The above verifies that DUFN can improve the rain removal performance by enhancing the ability to capture high-frequency information.

WT-U-Net++： surface defect detection network based on wavelet transform

Guohuan HE, Jiangping ZHU

2023, 43(10):  3260-3266.  DOI: 10.11772/j.issn.1001-9081.2022091452

Asbtract ( )   HTML ( )   PDF (2792KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To address the problems of traditional machine vision algorithms such as low detection accuracy， inability to adapt to environmental changes and noise influence in surface defect detection， a improved UNet++ based on Wavelet Transform （WT） — WT-U-Net++ was proposed. Firstly， the high frequency and low frequency components of the defect image were obtained by the WT， and the detailed features of the high and low frequency components were extracted by the multi-scale module MCI （Mix-Conv Inception）. Secondly， the detailed features extracted by MCI module were fused with the original image， and the fusion results were used as the input of the improved UNet++. Thirdly， in the downsampling stage of UNet++， channel attention module was introduced to enable the network to capture more contextual semantic information and improve the quality of cross-layer feature cascade at the same time. In the upsampling stage， deconvolution was adopted to recover more defect details. Finally， the best result was selected from the multiple output of UNet++ as the detection result. Experimental results on three public defect datasets of rail， magnetic tile and silicon steel oil stain show that compared with the sub-optimal algorithm UNet++， WT-U-Net ++ has the Intersection over Union （IoU） increased by 7.98%， 4.63%， and 8.74% respectively， and the Dice Similarity Coefficient （DSC） improved by 4.26%， 2.99% and 4.64% respectively.

Long-tailed image defect detection based on gradient-guide weighted-deferred negative gradient decay loss

Wei LI, Sixin LIANG, Jianzhou ZHANG

2023, 43(10):  3267-3274.  DOI: 10.11772/j.issn.1001-9081.2022091413

Asbtract ( )   HTML ( )   PDF (2403KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that the current image defect detection models have poor detection effect on tail categories in long-tail defect datasets， a GGW-DND Loss （Gradient-Guide Weighted-Deferred Negative Gradient decay Loss） was proposed. First， the positive and negative gradients were re-weighted according to the cumulative gradient ratio of the classification nodes in the detector in order to reduce the suppressed state of tail classifier. Then， once the model was optimized to a certain stage， the negative gradient generated by each node was sharply reduced to enhance the generalization ability of the tail classifier. Experimental results on the self-made image defect dataset and NEU-DET （NEU surface defect database for Defect dEtection Task） show that the mean Average Precision （mAP） for tail categories of the proposed loss is better than that of Binary Cross Entropy Loss （BCE Loss）， the former is increased by 32.02 and 7.40 percentage points respectively， and compared with EQL v2 （EQualization Loss v2）， the proposed loss has the mAP increased by 2.20 and 0.82 percentage points respectively， verifying that the proposed loss can effectively improve the detection performance of the network for tail categories.

Multi-scale feature enhanced retinal vessel segmentation algorithm based on U-Net

Zhiang ZHANG, Guangzhong LIAO

2023, 43(10):  3275-3281.  DOI: 10.11772/j.issn.1001-9081.2022091437

Asbtract ( )   HTML ( )   PDF (2624KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the shortcomings of traditional retinal vessel segmentation algorithm such as low accuracy of vessel segmentation and mis-segmentation of focal areas， a Multi-scale Feature Enhanced retinal vessel segmentation algorithm based on U-Net （MFEU-Net） was proposed. Firstly， in order to solve the vanishing gradient problem， an improved Feature Information Enhancement Residual Module （FIE-RM） was designed to replace the convolution block of U-Net. Secondly， in order to enlarge the receptive field and improve the extraction ability of vascular information features， a multi-scale dense atrous convolution module was introduced at the bottom of U-Net. Finally， in order to reduce the information loss in the process of encoding and decoding， a multi-scale channel enhancement module was constructed at the skip connection of U-Net. Experimental results on Digital Retinal Images for Vessel Extraction （DRIVE） and CHASE_DB1 datasets show that compared with CS-Net （Channel and Spatial attention Network）， the suboptimal algorithm in retinal vessel segmentation， MFEU-Net has the F1 score improved by 0.35 and 1.55 percentage points respectively， and the Area Under Curve （AUC） improved by 0.34 and 1.50 percentage points respectively. It is verified that MFEU-Net can improve the accuracy and robustness of retinal vessel segmentation effectively.

Multiscale dense fusion network for lung lesion image segmentation

Xiaoyan LU, Yang XU, Wenhao YUAN

2023, 43(10):  3282-3289.  DOI: 10.11772/j.issn.1001-9081.2022101545

Asbtract ( )   HTML ( )   PDF (3560KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems of incomplete segmentation of lung lesions and fuzzy prediction of regional boundaries in mainstream deep learning networks， a Multiscale Dense Fusion Network （MDF-Net） based on U-Net was proposed. Firstly， multi-branch dense skip connections were introduced to capture multi-level contextual information， and Information Weighted Fusion （IWF） module was introduced at the end of the network for level-by-level fusion to solve the feature loss problem in the network. Secondly， a self-attention pyramid module was designed. Each pyramid layer was used to segment the feature map in different scales， and the self-attention mechanism was applied to calculate the pixel correlation， thereby enhancing the saliency of the infection features in local and global regions. Finally， unlike the up-sampling form in traditional U-Net， a Up-sampling Residual （UR） module was designed. The multi-branch residual structure and channel feature excitation were used to help the network restore more abundant features of micro lesions. Experimental results on two public datasets show that compared with UNeXt， the proposed network improves the ACCuracy （ACC） by 1.5% and 1.4% respectively， and the Mean Intersection over Union （MIoU） by 3.9% and 1.9% respectively， which verify that MDF-Net has better lung lesion segmentation performance.

Frontier and comprehensive applications

UAV path planning for persistent monitoring based on value function iteration

Chen LIU, Yang CHEN, Hao FU

2023, 43(10):  3290-3296.  DOI: 10.11772/j.issn.1001-9081.2022091464

Asbtract ( )   HTML ( )   PDF (2422KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The use of Unmanned Aerial Vehicle （UAV） to continuously monitor designated areas can play a role in deterring invasion and damage as well as discovering abnormalities in time， but the fixed monitoring rules are easy to be discovered by the invaders. Therefore， it is necessary to design a random algorithm for UAV flight path. In view of the above problem， a UAV persistent monitoring path planning algorithm based on Value Function Iteration （VFI） was proposed. Firstly， the state of the monitoring target point was selected reasonably， and the remaining time of each monitoring node was analyzed. Secondly， the value function of the corresponding state of this monitoring target point was constructed by combining the reward/penalty benefit and the path security constraint. In the process of the VFI algorithm， the next node was selected randomly based on ε principle and roulette selection. Finally， with the goal that the growth of the value function of all states tends to be saturated， the UAV persistent monitoring path was solved. Simulation results show that the proposed algorithm has the obtained information entropy of 0.905 0， and the VFI running time of 0.363 7 s. Compared with the traditional Ant Colony Optimization （ACO）， the proposed algorithm has the information entropy increased by 216%， and the running time decreased by 59%，both randomness and rapidity have been improved. It is verified that random UAV flight path is of great significance to improve the efficiency of persistent monitoring.

Review on advances in recognition and classification of cognitive impairment based on EEG signals

Junpeng ZHANG, Yujie SHI, Rui JANG, Jingjing DONG, Changjian QIU

2023, 43(10):  3297-3308.  DOI: 10.11772/j.issn.1001-9081.2022101471

Asbtract ( )   HTML ( )   PDF (1199KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Early detection and timely intervention of cognitive impairment are crucial to slow down the progress of the disease. The ElectroEncephaloGraphy （EEG） signal has become an important tool for the investigation of biomarkers of cognitive diseases due to its high temporal resolution and easy acquisition. Compared with the traditional biomarker recognition method， the machine learning method has higher accuracy and better stability for the recognition and classification of cognitive impairment based on EEG signals. Aiming at the relevant research literature on the recognition and classification of cognitive impairment based on EEG signals in the past three years， firstly， from the perspectives of five categories of EEG features commonly used in the recognition and classification of cognitive impairment， including time domain， frequency domain， combination of time and frequency domains， nonlinear dynamics， functional connectivity and brain network， more representative EEG features were found. Then， the currently commonly used classification methods based on machine learning and deep learning， such as Support Vector Machine （SVM）， Linear Discriminant Analysis （LDA）， K-Nearest Neighbor （KNN） and Artificial Neural Network （ANN）， as well as their performance were summarized. Finally， the current problems in different kinds of studies were analyzed， and the future research directions in this field were prospected， thereby providing reference for the follow-up research on the recognition and classification of cognitive impairment based on EEG signals.

Gene splice site identification based on BERT and CNN

Min ZUO, Hong WANG, Wenjing YAN, Qingchuan ZHANG

2023, 43(10):  3309-3314.  DOI: 10.11772/j.issn.1001-9081.2022091447

Asbtract ( )   HTML ( )   PDF (1829KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the development of high-throughput sequencing technology， massive genome sequence data provide a data basis to understand the structure of genome. As an essential part of genomics research， splice site identification plays a vital role in gene discovery and determination of gene structure， and is of great importance for understanding the expression of gene traits. To address the problem that existing models cannot extract high-dimensional features of DNA （DeoxyriboNucleic Acid） sequences sufficiently， a splice site prediction model consisted of BERT （Bidirectional Encoder Representations from Transformers） and parallel Convolutional Neural Network （CNN） was constructed， namely BERT-splice. Firstly， the DNA language model was trained by BERT pre-training method to extract the contextual dynamic association features of DNA sequences and map DNA sequence features with a high-dimensional matrix. Then， the DNA language model was used to map the human reference genome sequence hg19 data into a high-dimensional matrix， and the result was adopted as input of parallel CNN classifier for retraining. Finally， a splice site prediction model was constructed on the basis of the above. Experimental results show that the prediction accuracy of BERT-splice model is 96.55% on the donor set of DNA splice sites and 95.80% on the acceptor set， which improved by 1.55% and 1.72% respectively， compared to that of the BERT and Recurrent Convolutional Neural Network （RCNN） constructed prediction model BERT-RCNN. Meanwhile， the average False Positive Rate （FPR） of donor/acceptor splice sites tested on five complete human gene sequences is 4.74%. The above verifies that the effectiveness of BERT-splice model for gene splice site prediction.