Table of Content

10 September 2022, Volume 42 Issue 9

Previous Issue   

Artificial intelligence

Semi-supervised representation learning method combining graph auto-encoder and clustering

Hangyuan DU, Sicong HAO, Wenjian WANG

2022, 42(9):  2643-2651.  DOI: 10.11772/j.issn.1001-9081.2021071354

Asbtract ( )   HTML ( )   PDF (1000KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Node label is widely existed supervision information in complex networks， and it plays an important role in network representation learning. Based on this fact， a Semi-supervised Representation Learning method combining Graph Auto-Encoder and Clustering （GAECSRL） was proposed. Firstly， the Graph Convolutional Network （GCN） and inner product function were used as the encoder and the decoder respectively， and the graph auto-encoder was constructed to form an information dissemination framework. Then， the k-means clustering module was added to the low-dimensional representation generated by the encoder， so that the training process of the graph auto-encoder and the category classification of the nodes were used to form a self-supervised mechanism. Finally， the category classification of the low-dimensional representation of the network was guided by using the discriminant information of the node labels. The network representation generation， category classification， and the training of the graph auto-encoder were built into a unified optimization model， and an effective network representation result that integrates node label information was obtained. In the simulation experiment， the GAECSRL method was used for node classification and link prediction tasks. Experimental results show that compared with DeepWalk， node2vec， learning Graph Representations with global structural information （GraRep）， Structural Deep Network Embedding （SDNE） and Planetoid （Predicting labels and neighbors with embeddings transductively or inductively from data）， GAECSRL has the Micro?F1 index increased by 0.9 to 24.46 percentage points， and the Macro?F1 index increased by 0.76 to 24.20 percentage points in the node classification task； in the link prediction task， GAECSRL has the AUC （Area under Curve） index increased by 0.33 to 9.06 percentage points， indicating that the network representation results obtained by GAECSRL effectively improve the performance of node classification and link prediction tasks.

Model distillation model based on training weak teacher networks about few-shots

Chunhao CAI, Jianliang LI

2022, 42(9):  2652-2658.  DOI: 10.11772/j.issn.1001-9081.2021071201

Asbtract ( )   HTML ( )   PDF (1828KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the lack of training data of deep neural networks in image recognition， as well as the loss of detailed features and the large amount of distillation calculations in the multi-model distillation， a model distillation model based on training weak teacher networks about few-shots was proposed. Firstly， the weak teacher network set was trained through the Bootstrap aggregating （Bagging） algorithm in the ensemble learning algorithm. While retaining the detailed features of the image dataset， parallel computing was able to be realized to improve the efficiency of network generation. Then， the knowledge merging algorithm was combined， and single high-quality high-complexity teacher networks were formed based on the weak teacher network feature maps， thereby obtaining the image feature maps with more significant details. Finally， based on the current advanced model distillation， an ensemble distillation algorithm with meta-network improved with combined feature maps was proposed， which reduced the calculation of meta-network training and realized the training of the target network about few-shots at the same time. Experimental results show that the algorithm had a 6.39% relative improvement in accuracy compared to the distillation scheme that uses a high-quality network as the teacher network. Comparing the accuracy of the model which obtained by training and distilling the teacher networks with Adaptive Boosting （AdaBoost） algorithm and the accuracy of the model obtained by the ensemble distillation model， the difference is within the given error range. However， the network generation rate of the ensemble distillation algorithm was increased by 4.76 times compared with that of AdaBoost algorithm. Therefore， the proposed algorithm can effectively improve the accuracy and training efficiency of the target model about few-shots.

Real-time segmentation algorithm based on attention mechanism and effective factorized convolution

Kai WEN, Weiwei TANG, Junchen XIONG

2022, 42(9):  2659-2666.  DOI: 10.11772/j.issn.1001-9081.2021071327

Asbtract ( )   HTML ( )   PDF (2344KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The current real-time semantic segmentation algorithm has the high computational cost and large memory footprint， which cannot meet the applications requirements of actual scenes. In order to solve the problems， a new type of shallow lightweight real-time semantic segmentation algorithm — AEFNet （Real-time segmentation algorithm based on Attention mechanism and Effective Factorized convolution） was proposed. Firstly， one-dimensional non-bottleneck structure （Non-bottleneck-1D） was adopted to construct a lightweight factorized convolution module to extract rich contextual information and reduce the amount of calculation. At the same time， the learning ability of the algorithm was enhanced in a simple way and the extraction of detailed information was facilitated. Then， the pooling operation and Attention Refinement Module （ARM） were combined to construct a global context attention module to capture global information and refine each stage of the algorithm to optimize the segmentation effect. The algorithm was verified on the public datasets cityscapes and camvid， and the precision of 74.0% and the inference speed of 118.9 Frames Per Second （FPS） were obtained on the cityscapes test set. Compared with Depth-wise Asymmetric Bottleneck Network （DABNet）， the proposed algorithm has the precision increased by about 4 percentage points， and the inference speed increased by 14.7 FPS. Compared with the recent efficient Enhanced Asymmetric Convolution Network （EACNet）， the proposed algorithm has the precision slightly lower by 0.2 percentage points， but has the inference speed increased by 6.9 FPS. Experimental results show that the proposed algorithm can more accurately identify the scene information， and can meet the real-time requirements.

Key event extraction method from microblog by integrating social influence and temporal distribution

Xujian ZHAO, Chongwei WANG, Junli WANG

2022, 42(9):  2667-2673.  DOI: 10.11772/j.issn.1001-9081.2021071330

Asbtract ( )   HTML ( )   PDF (2009KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that the existing microblog event extraction methods are based on the content characteristics of events and ignore the relationship between the social attributes and time characteristics of events， so that they cannot identify the key events in the propagation process of microblog hot spots， a key event extraction method from microblog by integrating social influence and temporal distribution was proposed. Firstly， the social influence was modeled to present importance of microblog events. Secondly， the temporal characteristics of microblog events during evolution were considered to capture the differences of events under different temporal distributions. Finally， the key microblog events were extracted under different temporal distributions. Experimental results on real datasets show that the proposed method can effectively extract key events in microblog hot spots. Compared with four methods of random selection， Term Frequency-Inverse Document Frequency （TF-IDF）， minimum-weight connected dominating set and degree and clustering coefficient information， the proposed method has the event set integrity index improved by 21%， 18%， 26% and 30% on dataset 1 respectively， and 14%， 2%， 21% and 23% on dataset 2 respectively. The extraction effect of the proposed method is better than those of the traditional methods.

Attention sentiment analysis model based on multi-scale convolution and gating mechanism

Hongjun HENG, Tianbao XU

2022, 42(9):  2674-2679.  DOI: 10.11772/j.issn.1001-9081.2021081448

Asbtract ( )   HTML ( )   PDF (682KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that most of existing models for document-level sentiment analysis only consider encoding text at word level， an attention sentiment analysis model based on multi-scale convolution and gating mechanism was proposed. Firstly， in order to obtain more different levels of text semantic information and form a richer text representation， the multi-scale convolution was used to capture local correlations of different granularities. Secondly， considering the influence of user personality and product information on text sentiment classification， the global information of users and products was integrated into attention to capture key semantic components which were highly related to users and products to form the document representation. Thirdly， a gating mechanism was introduced to control the path of emotional information flowing to collection layer. Finally， the sentiment classification was realized through the fully connected layer and argmax function. The experimental results show that， compared with the baseline model with the most advanced performance， the proposed algorithm has the sentiment classification accuracy on IMDB and Yelp2014 datasets improved by 1.2 percentage points and 0.7 percentage points respectively， and obtains the smallest Root Mean Squared Error （RMSE） on IMDB and Yelp2013 datasets.

Chinese named entity recognition based on knowledge base entity enhanced BERT model

Jie HU, Yan HU, Mengchi LIU, Yan ZHANG

2022, 42(9):  2680-2685.  DOI: 10.11772/j.issn.1001-9081.2021071209

Asbtract ( )   HTML ( )   PDF (1391KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that the pre-training model BERT （Bidirectional Encoder Representation from Transformers） lacks of vocabulary information， a Chinese named entity recognition model called OpenKG + Entity Enhanced BERT + CRF （Conditional Random Field） based on knowledge base entity enhanced BERT model was proposed on the basis of the semi-supervised entity enhanced minimum mean-square error pre-training model. Firstly， documents were downloaded from Chinese general encyclopedia knowledge base CN-DBPedia and entities were extracted by Jieba Chinese text segmentation to expand entity dictionary. Then， the entities in the dictionary were embedded into BERT for pre-training. And the word vectors obtained from the training were input into Bidirectional Long-Short-Term Memory network （BiLSTM） for feature extraction. Finally， the results were corrected by CRF and output. Model validation was performed on datasets CLUENER 2020 and MSRA， and the proposed model was compared with Entity Enhanced BERT pre-training， BERT+BiLSTM， ERNIE and BiLSTM+CRF models. Experimental results show that compared with these four models， the proposed model has the F1 score increased by 1.63 percentage points and 1.1 percentage points， 3.93 percentage points and 5.35 percentage points， 2.42 percentage points and 4.63 percentage points， 6.79 and 7.55 percentage points， respectively in the two datasets. It can be seen that the comprehensive effect of the proposed model on named entity recognition is effectively improved， and the F1 scores of the model are better than those of the comparison models.

Medical named entity recognition model based on deep auto-encoding

Xudong HOU, Fei TENG, Yi ZHANG

2022, 42(9):  2686-2692.  DOI: 10.11772/j.issn.1001-9081.2021071317

Asbtract ( )   HTML ( )   PDF (979KB) ( )  

Figures and Tables | References | Related Articles | Metrics

With the deepening of the network in the Medical Named Entity Recognition （MNER） problem， the recognition accuracy and computing power requirements of the deep learning-based recognition models are unbalanced. Aiming at this problem， a medical named entity recognition model CasSAttMNER （Cascade Self-Attention Medical Named Entity Recognition） based on deep auto-encoding was proposed. Firstly， a depth difference balance strategy between encoding and decoding was used in the model， and the distilled Transformer language model RBT6 was used as the encoder to reduce the encoding depth and the computing power requirements for training and application. Then， Bidirectional Long Short-Term Memory （BiLSTM） network and Conditional Random Field （CRF） were used to propose a cascaded multi-task dual decoder to complete entity mention sequence labeling and entity class determination. Finally， based on the self-attention mechanism， the model design was optimized by effectively representing the implicit decoding information between the entity classes and the entity mentions. Experimental results show that the F value measurements of CasSAttMNER on two Chinese medical entity datasets can reach 0.943 9 and 0.945 7， which are 3 percentage points and 8 percentage points higher than those of the baseline model， respectively， verifying that this model further improves the decoder performance.

Python named entity recognition model based on transformer

Guanyou XU, Weisen FENG

2022, 42(9):  2693-2700.  DOI: 10.11772/j.issn.1001-9081.2021071356

Asbtract ( )   HTML ( )   PDF (1723KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Recently， some character-based Named Entity Recognition （NER） models cannot make full use of word information， and the lattice structure model using word information may degenerate into a word-based model and cause word segmentation errors. To deal with these problems， a python NER model based on transformer was proposed to encode character-word information. Firstly， the word information was bound to the characters corresponding to the beginning or end of the word. Then， three different strategies were used to encode the word information into a fixed-size representation through the transformer. Finally， Conditional Random Field （CRF） was used for decoding， thereby avoiding the problem of word segmentation errors caused by obtaining the word boundary information as well as improving the batch training speed. Experimental results on the python dataset show that the F1 score of the proposed model is 2.64 percentage points higher than that of the Lattice-LSTM model， and the training time of the proposed model is about a quarter of the comparison model， indicating that the proposed model can prevent model degradation， improve batch training speed， and better recognize the python named entities.

Data science and technology

Attribute reduction algorithm based on cluster granulation and divergence among clusters

Yan LI, Bin FAN, Jie GUO

2022, 42(9):  2701-2712.  DOI: 10.11772/j.issn.1001-9081.2021081371

Asbtract ( )   HTML ( )   PDF (3592KB) ( )   PDF（mobile） (654KB) ( 12 )  

Figures and Tables | References | Related Articles | Metrics

Attribute reduction is a hot research topic in rough set theory. Most of the algorithms of attribute reduction for continuous data are based on dominance relations or neighborhood relations. However， continuous datasets do not necessarily have dominance relations in attributes. And the attribute reduction algorithms based on neighborhood relations can adjust the granulation degree through neighborhood radius， but it is difficult to unify the radii due to the different dimensions of attributes and the continuous values of radius parameters， resulting in high computational cost of the whole parameter granulation process. To solve this problem， a multi-granularity attribute reduction strategy based on cluster granulation was proposed. Firstly， the similar samples were classified by the clustering method， and the concepts of approximate set， relative positive region and positive region reduction based on clustering were proposed. Secondly， according to JS （Jensen-Shannon） divergence theory， the difference of data distribution of each attribute among clusters was measured， and representative features were selected to distinguish different clusters. Finally， an attribute reduction algorithm was designed using a discernibility matrix. In the proposed algorithm， the attributes were not required to have ordered relations. Different from neighborhood radius， the clustering parameter was discrete， and the dataset was able to be divided into different granulation degrees by adjusting this parameter. Experimental results on UCI and Kent Ridge datasets show that this attribute reduction algorithm can directly deal with continuous data. At the same time， by using this algorithm， the redundant features in the datasets can be removed while maintaining or even improving the classification accuracy by discrete adjustment of the parameters in a small range.

Statistically significant sequential patterns mining algorithm under influence degree

Jun WU, Aijia OUYANG, Lin ZHANG

2022, 42(9):  2713-2721.  DOI: 10.11772/j.issn.1001-9081.2021071311

Asbtract ( )   HTML ( )   PDF (1068KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems that the degree of support is not a good indicator for the interestingness of sequential patterns and the quality of reported sequential patterns is not evaluated in traditional sequential patterns mining algorithms， a statistically significant sequential patterns mining algorithm under influence degree， calling ISSPM （Influence-based Significant Sequential Patterns Mining）， was proposed. Firstly， all sequential patterns meeting the interestingness constraint were mined recursively. Then， the itemset permuting method was introduced to construct permutation test null distribution for these sequential patterns. Finally， the statistical measures of the evaluated sequential patterns were calculated from this distribution， and all statistically significant sequential patterns were found from the above sequential patterns. In the experiments with the PSPM （Prefix-projected Sequential Patterns Mining）， SPDL （Sequential Patterns Discovering under Leverage） and PSDSP （Permutation Strategies for Discovering Sequential Patterns） algorithms on the real-world sequential record datasets， ISSPM algorithm reports fewer but more interesting sequential patterns. Experimental results on the synthetic sequential record datasets show that the average proportion of the false positive sequential patterns reported by the ISSPM algorithm is 3.39%， and the discovery rate of embedded patterns of this algorithm is not less than 66.7%， which are significantly better than those of the above three algorithms to compare. It can be seen that the statistically significant sequential patterns reported by ISSPM algorithm can reflect more valuable information in sequential record datasets， and the decisions made based on the information are more reliable.

Predictability evaluation and joint forecasting method for intermittent time series

Yiping LANG, Wentao MAO, Tiejun LUO, Lilin FAN, Yingying REN, Xia LIU

2022, 42(9):  2722-2731.  DOI: 10.11772/j.issn.1001-9081.2021071196

Asbtract ( )   HTML ( )   PDF (4732KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In the operation and maintenance of high-end manufacturing enterprises， the spare parts demand occurs randomly， accompanied by a large number of zero demand periods. At the same time， the corresponding sparse parts demand data is of small scale and has intermittent and distribution with lump formation characteristics. Consequently， most of current time series forecasting methods are hard to effectively predict the demand trends. To solve this problem， a predictability evaluation and joint forecasting method for intermittent time series was proposed. Firstly， a new intermittent-similarity metric was proposed. In this metric， the frequency and positions of the "0" element occurring in the two sequences were counted， while the metrics such as maximal information coefficient and average demand interval were combined to evaluate the tendency information and fluctuation pattern of the sequences effectively and realize the quantification of the predictability of the intermittent time series. Then， based on this metric， an intermittent-similarity hierarchical clustering method was constructed to adaptively select the sequences with high similarity and strong predictability as well as eliminate extremely sparse and unpredictable sequences. Moreover， the structured information between the sequences was explored and utilized， a Multi-output Support Vector Regression （M-SVR） model was constructed， thereby achieving the joint prediction of intermittent time series with small-scale data. Finally， the experiments were conducted on two public datasets （UCI （University of California Irvine） gift retail dataset and Huawei computer accessory dataset） and a real-world spare parts after-sales dataset of a large manufacturing enterprise， respectively. The results show that compared with several representative time series forecasting methods， the proposed method can effectively exploit the predictability of various kinds of intermittent sequences and improve the prediction accuracy of intermittent time series with small-scale data. Therefore， the proposed method provides a new solution for the spare parts demand forecasting of manufacturing enterprises.

Cyber security

Review of white-box adversarial attack technologies in image classification

Jiaxuan WEI, Shikang DU, Zhixuan YU, Ruisheng ZHANG

2022, 42(9):  2732-2741.  DOI: 10.11772/j.issn.1001-9081.2021071339

Asbtract ( )   HTML ( )   PDF (2101KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In the research of image classification tasks in deep learning， the phenomenon of adversarial attacks brings severe challenges to the secure application of deep learning models， which arouses widespread attention of researchers. Firstly， around the adversarial attack technologies for generating the adversarial perturbations， the important white-box adversarial attack algorithms in the image classification tasks were introduced in detail， and the advantages and disadvantages of different attack algorithms were analyzed. Then， from three realistic application scenarios： mobile application， face recognition and autonomous driving， the application status of the white-box adversarial attack technologies was illustrated. Additionally， some typical white-box adversarial attack algorithms were selected to perform experiments on different target models， and the experimental results were analyzed. Finally， the white-box adversarial attack technologies were summarized， and their valuable research directions were prospected.

Blow-CAST-Fish key recovery attack based on differential tables

Xiaoling SUN, Shanshan LI, Guang YANG, Qiuge YANG

2022, 42(9):  2742-2749.  DOI: 10.11772/j.issn.1001-9081.2021071340

Asbtract ( )   HTML ( )   PDF (1646KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems of limited attack rounds and high attack complexity of Blow-CAST-Fish （Blow-C.Adams S.Tavares-Fish） algorithm， a key recovery attack of Blow-CAST-Fish algorithm based on differential table was proposed. Firstly， after analyzing the collision of S-box， based on the collision of two S-boxes and a single S-box respectively， the 6-round and 12-round differential characteristics were constructed. Secondly， the differential tables of f₃ were calculated， and three rounds were expanded based on the specific differential characteristic， thereby determining the relationship between ciphertext difference and the input and output differences of f₃. Finally， the plaintexts meeting the conditions were selected to encrypt， the input and output differences of f₃ were calculated according to the ciphertext difference， and the corresponding input and output pairs were found by querying the differential table， as a result， the subkeys were obtained. At the situation of two S-boxes collision， the proposed attack completed a differential attack of 9-round Blow-CAST-Fish algorithm， compared with the comparison attack， the number of attack rounds was increased by one， and the time complexity was reduced from 2^107.9 to 2⁷⁴. At the situation of single S-box collision， the proposed attack completed a differential attack of 15-round Blow-CAST-Fish algorithm， compared with the comparison attack， although the number of attack rounds was reduced by one， the proportion of weak keys was increased from {\mathrm{2}}^{-\mathrm{52.4}} to {\mathrm{2}}^{-\mathrm{42}} and the data complexity was reduced from 2⁵⁴ to 2⁴⁷. The test results show that the attack based on differential table can increase the efficiency of attack based on the same differential characteristics.

GPU-based method for evaluating algebraic properties of cryptographic S-boxes

Jingwen CAI, Yongzhuang WEI, Zhenghong LIU

2022, 42(9):  2750-2756.  DOI: 10.11772/j.issn.1001-9081.2021081382

Asbtract ( )   HTML ( )   PDF (2206KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Cryptographic S-boxes （or black boxes） are nonlinear components in symmetric encryption algorithms， and their algebraic properties usually determine the security performance of these encryption algorithms. Differential uniformity， nonlinearity and revised transparency order are three basic indicators to evaluate the security properties of cryptographic S-boxes. They describe the S-box’s ability against differential cryptanalysis， linear cryptanalysis and differential power attack respectively. When the input size of the cryptographic S-box is large （for example， the input length of the S-box is larger than 15 bits）， the needed solving time in Central Processing Unit （CPU） is still too long， or even the solution is impracticable. How to evaluate the algebraic properties of the large-size S-box quickly is currently a research hot point in the field. Therefore， a method to evaluate the algebraic properties of cryptographic S-boxes quickly was proposed on the basis of Graphics Processing Unit （GPU）. In this method， the kernel functions were split into multiple threads by slicing technique， and an optimization scheme was proposed by combining the characteristics of solving differential uniformity， nonlinearity and revised transparency order to realize parallel computing. Experimental results show that compared with CPU-based implementation environment， single GPU based environment has the implementation efficiency significantly improved. Specifically， the time spent on calculating differential uniformity， nonlinearity， and revised transparency order is saved by 90.28%， 80%， and 66.67% respectively， which verifies the effectiveness of this method.

Supervisable blockchain anonymous transaction system model

Yangnan GUO, Wenbao JIANG, Shuai YE

2022, 42(9):  2757-2764.  DOI: 10.11772/j.issn.1001-9081.2021081408

Asbtract ( )   HTML ( )   PDF (2009KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In order to solve the problem that the traceability and privacy protection of the existing blockchain systems are difficult to balance， a supervisable blockchain anonymous transaction system model was designed. Firstly， the advantages of Identity-Based Cryptograph （IBC） and Certificateless Public Key Cryptography （CL-PKC） were combined to remove the hidden dangers caused by the single IBC key escrow， and the user transaction identification was associated with the user identity under safe conditions， thereby ensuring the supervisability of the model. Then， privacy security was achieved through double-layer signature authentication from the network layer and the application layer， which not only guaranteed the security of users’ transaction content and identity privacy， but also ensured that the authority was able to trace back based on abnormal transactions， providing a new idea for the compatibility between the current blockchain anonymity and supervisability. Finally， the proposed model was compared with the self-certified signature model， the multi-center SM9 model and the traceable Monero model， and the transmission time consumptions of this model and the mainstream blockchain systems were compared by computer simulation. Experimental results show that the proposed model has greater advantages in security and traceability； under the same hardware and software environment， the proposed model consumes 168% more time than the Ethereum model when transmitting the same size information for several times， and the difference in efficiency is not significant in the case of long-time transmission； the proposed model consumes 38% more time than the Ethereum model on average when transmitting information of different lengths.

Data trusted traceability method based on Merkle mountain range

Wei LIU, Cong ZHANG, Wei SHE, Xuan SONG, Zhao TIAN

2022, 42(9):  2765-2771.  DOI: 10.11772/j.issn.1001-9081.2021081369

Asbtract ( )   HTML ( )   PDF (2701KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Concerning the problems of the high cost of massive data storage and low efficiency of data traceability verification in the Internet of Things （IoT） system， a data trusted traceability method based on Merkel Mountain Range （MMR）， named MMRBCV （Merkle Mountain Range BlockChain Verification）， was proposed. Firstly， Inter-Planetary File System （IPFS） was used to realize the storage of the IoT data. Secondly， the consortium blockchains and private blockchains were adopted to design a double-blockchain structure to realize reliable recording of the data flow process. Finally， based on the MMR， a block structure was constructed to realize the rapid verification of lightweight IoT nodes in the process of data traceability. Experimental results show that MMRBCV reduces the amount of data downloaded during data tracing， and the data verification time is related to the structure of MMR. When MMR forms a perfect binary tree， the data verification time is short. When the block height is 200 000， MMRBCV’s maximum verification time is about 10 ms， which is about 72% shorter than that of Simplified Payment Verification （SPV） （about 36 ms）， indicating that the proposed method improves the verification efficiency effectively.

Attribute-based encryption scheme with verifiable search and non-monotonic access structure

Suqing LIN, Shuhua ZHANG

2022, 42(9):  2772-2779.  DOI: 10.11772/j.issn.1001-9081.2021081446

Asbtract ( )   HTML ( )   PDF (975KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Most existing attribute-based keyword search schemes only support monotonic access structure and lack efficient verification for search results. Aiming at these problems， a ciphertext keyword search attribute-based encryption scheme with verifiable search and non-monotonic access structure was proposed. Firstly， the polynomials were constructed by the attribute values， and the fine-grained ciphertext search permission setting was accomplished by divisibility property of the polynomials. Then， both keyword search and outsourced decryption were performed by the cloud servicer without revealing any private information. Finally， the search result verification was realized by utilizing the proposed commitment scheme. The proposed scheme supports multiple functions such as non-monotonic access structure， fine-grained search， data sharing， outsourced decryption， and verifiable search. Under the augmented Multi-Sequence of Exponents Decisional Diffie-Hellman （aMSE-DDH） assumption， it can be proved that this scheme has selective indistinguishability security under chosen ciphertext attacks and under chosen keyword attacks， respectively， in the random oracle model. Experimental results show that the terminal decryption time of the proposed scheme is not related to the attribute number， and is about 12.9 ms.

Moving target defense decision-making algorithm based on multi-stage evolutionary signal game model

Wenting BI, Haitao LIN, Liqun ZHANG

2022, 42(9):  2780-2787.  DOI: 10.11772/j.issn.1001-9081.2021071154

Asbtract ( )   HTML ( )   PDF (2390KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Currently， the network security accidents occur frequently， and traditional passive defense technologies have no possible response to unknown network security threats. In response to this problem， a multi-stage evolutionary signal game model was constructed. And with the background that the defender actively launches inductive signals for security defense， a Moving Target Defense （MTD） decision-making algorithm based on the multi-stage evolutionary signal game model was proposed. Firstly， the basic elements of the model were defined and the overall model was analyzed theoretically based on the assumptions of incomplete information constraints and complete rationality of both sides of the game. Then， a method for quantifying the benefits of offensive and defensive strategies was designed， and a detailed optimal strategy solving process for equilibrium was given. Finally， the MTD method was introduced to analyze the evolution trends of both sides’ strategies in multi-stage attack and defense. Experimental results show that the proposed algorithm can predict the optimal defense strategies at different stages accurately， and has guiding significance for the research of new network active defense technology. At the same time， the results of comparing the proposed algorithm with the traditional random uniform strategy selection algorithm through Monte Carlo simulation experiment verify the effectiveness and safety of the proposed algorithm.

Advanced computing

Yin-Yang-pair optimization algorithm based on dynamic D-way splitting and chaotic perturbation

Dahai LI, Qingteng LIU, Zhigang AI, Zhendong WANG

2022, 42(9):  2788-2799.  DOI: 10.11772/j.issn.1001-9081.2021071342

Asbtract ( )   HTML ( )   PDF (1671KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To improve the performance of Yin-Yang-Pair Optimization-Simulated Annealing1 （YYPO-SA1）， a Yin-Yang-pair optimization algorithm based on dynamic D-way splitting and chaotic perturbation NYYPO （Newton-Yin-Yang-Pair Optimization） was proposed. Firstly， in order to dynamically adjust the probability of D-way splitting， Newton’s law of cooling mechanism was adopted. Then， the chaotic perturbation strategy was applied in splitting stage. The dynamic adjustment mechanism was applied to enable NYYPO to use a larger D-way segmentation probability at the early stage of search， and use a smaller D-way segmentation probability at the late stage of search， which enhanced the global search ability of the algorithm. Meanwhile， the diversity of solution was enriched， and the ability of the algorithm to jump out of local optimum was improved by using chaotic perturbation strategy. Finally， NYYPO was applied to the parameter optimization design problem of wind-driven generator. Fifteen test functions， including unimodal， multimodal， and composite functions， were selected to evaluate the performance of NYYPO， YYPO-SA1， and 6 representative single-objective optimization algorithms： Particle Swarm Optimization （PSO） algorithm， Crow Search Algorithm （CSA）， Grey Wolf Optimizer （GWO）， Whale Optimization Algorithm （WOA）， Flower Pollination Algorithm （FPA）， and Sparrow Search Algorithm （SSA）. The results show that compared with YYPO-SA1， NYYPO obtains 12 orders of magnitude improvement on Sphere function. In Friedman test， when dimension is 10， 30， 50 respectively， NYYPO ranks 2.87， 2.0 and 1.93 averagely and respectively， total ranking of all of them is the first. It can be seen that NYYPO achieves significant performance advantages in statistical significance. At the same time， NYYPO also achieves better optimization results in the parameter optimization design problem of wind-driven generator.

Pooling algorithm based on Gaussian function

Yuhang WANG, Yongxia ZHOU, Liangwu WU

2022, 42(9):  2800-2806.  DOI: 10.11772/j.issn.1001-9081.2021071216

Asbtract ( )   HTML ( )   PDF (1518KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that the traditional pooling algorithms in Convolutional Neural Network （CNN） cannot well consider the correlation between each element in the pooling domain and the features contained in the pooling domain， a pooling algorithm based on Gaussian function was proposed. Firstly， according to the value of each element in the pooling domain and the maximum value of all elements， the three parameter values of the Gaussian function were calculated. Then， the Gaussian function was used to calculate the weights of all elements in the pooling domain. Finally， the weighted average value of all elements in the pooling domain was calculated according to these weights. Finally， the obtained value was used as the pooling result. LeNet5， VGG （Visual Geometry Group）16， ResNet （Residual Network）18 and MobileNet v3 were selected as the experimental models. Experiments were carried out on public datasets CIFAR-10， Fer2013 and German Traffic Sign Recognition Benchmark （GTSRB）， and max pooling， average pooling， random pooling， mixed pooling， fuzzy pooling， fused random pooling and soft pooling were selected to compare. Experimental results show that the proposed algorithm improves the accuracy by 0.5 percentage points to 6 percentage points compared with other algorithms on the three datasets， and the running efficiency of the proposed algorithm is higher than those of the other pooling algorithms except max pooling algorithm and average pooling algorithm， so as to verify that the proposed algorithm is effective and suitable for the situations where the operation time demand is not high but the accuracy demand is high.

Archimedes optimization algorithm based on golden Levy guidance mechanism

Jun CHEN, Qing HE, Shouyu LI

2022, 42(9):  2807-2815.  DOI: 10.11772/j.issn.1001-9081.2021081438

Asbtract ( )   HTML ( )   PDF (1984KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems of standard Archimedes Optimization Algorithm （AOA） in solving optimization problems， such as weak global exploration ability， slow convergence and low solution accuracy， a Multi-Strategy improved AOA （MSAOA） was proposed. Firstly， the variable interval initialization strategy was used to make initial population near to the global optimal solution as close as possible to improve the quality of initial solution. Secondly， the golden Levy guidance mechanism was proposed to improve the population diversity of the algorithm in later iteration stage. Thirdly， the adaptive wavelength operator was introduced to achieve the purpose of improving search efficiency of the algorithm while maintaining diversity of population. The proposed algorithm was compared with Equilibrium Optimizer （EO）， Sine Cosine Algorithm （SCA） and Grey Wolf Optimizer （GWO） on 20 benchmark test functions. Experimental results show that the proposed algorithm has higher optimization accuracy and convergence speed. And the proposed algorithm was applied to four mechanical design examples to verify the effectiveness and superiority of the proposed algorithm again.

Network and communications

Buffer compensation based video bitrate adaptation algorithm

Aiguo JI, Yunzhe LUAN

2022, 42(9):  2816-2822.  DOI: 10.11772/j.issn.1001-9081.2021081394

Asbtract ( )   HTML ( )   PDF (2310KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Concerning the problem that the existing bitrate adaptation algorithms based on Dynamic Adaptive Streaming over Hyper Text Transfer Protocol （HTTP） （DASH） have low buffer utilization and low average bitrate， a Bitrate Adaptive Switching based on Buffer Compensation （BASBC） algorithm based on DASH standard was proposed. Firstly， the bandwidth fluctuation was analyzed according to the download speeds of the recent downloaded fragments， and thereby obtaining the estimated bandwidth. Secondly， according to the estimated bandwidth and current bitrate level， bitrate higher/lower switching thresholds were set in the buffer to control bitrate switching to higher level and consum buffer time， or to control bitrate switching down gradually and accumulate buffer time， respectively， so that a consumption-accumulation buffer state loop was formed in video buffer. The average bitrate of video playback of BASBC algorithm is higher than that of Dynamic Adaptive Step-wise Bitrate Switching （DASBS） algorithm for HTTP streaming， thereby effectively improving the bandwidth utilization. Although the average bitrate of the proposed algorithm is slightly lower than that of Rate Smooth Switching （RSS） algorithm based on DASH standard， the smoothness of bitrate switching of the proposed algorithm is better， improving the switching stability. Experimental results show that the proposed algorithm has good performance such as high bandwidth utilization， switching smoothness and switching stability in dynamic network environment， which can effectively improve the Quality of Experience （QoE） of users.

Link prediction algorithm based on information entropy improved PCA model

Yuyu MENG, Jing GUO

2022, 42(9):  2823-2829.  DOI: 10.11772/j.issn.1001-9081.2021071326

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

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that traditional link prediction has computational results not stable in networks with different structures， a link prediction algorithm based on information entropy improved Principal Component Analysis （PCA） model was proposed. Firstly， seven similarity indexes were determined by Random Forest （RF） as the optimal feature set. Then， seven similarity indexes were combined to propose a feature information fusion model based on information entropy improved PCA. After weighting the feature information， the model was combined with the single mechanism algorithms to verify the correctness and verification effect of the model on six real-world datasets. Finally， the feasibility and effectiveness of the link prediction algorithm based on the proposed model were verified by comparing Area Under the Curve （AUC） values with the hybrid link prediction algorithms. Experimental results show that the proposed link prediction algorithms improve the AUC value by 2.5 to 12.46 percentage points and 0.47 to 9.01 percentage points， respectively， compared with Ordered Weighted Averaging aggregation operator （OWA） and Ensemble-Model-based Link Prediction algorithm （EMLP）. It can be seen that applying the proposed algorithm to networks with different structural features can obtain more stable and accurate link prediction results.

Multimedia computing and computer simulation

Kinect-based human pose estimation optimization and animation generation

Wei ZHAO, Yi LI

2022, 42(9):  2830-2837.  DOI: 10.11772/j.issn.1001-9081.2021061043

Asbtract ( )   HTML ( )   PDF (7004KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In order to generate more accurate and smooth virtual human animation， the Kinect device was used to capture the 3D human body pose data， and the monocular 3D human body pose estimation algorithm was used to reason the skeleton points in the color information of the Kinect at the same time， thereby optimizing the human pose estimation effect at real time， and driving the virtual character model to generate animation. Firstly， a spatio-temporal optimization method of skeleton point data processing was proposed to improve the stability of monocular estimation of the 3D human body pose. Secondly， a human pose estimation method based on the fusion of Kinect and Occlusion-Robust Pose-Maps （ORPM） algorithm was proposed to solve the occlusion problem of Kinect. Finally， a virtual human animation system based on quaternion vector interpolation and inverse kinematics constraints was developed， which was able to perform motion simulation and real-time animation generation. Compared with the animation generation method that only uses Kinect to capture human motion， the proposed method has the human body estimation data more robust， and has a certain anti-occlusion ability. The animation frame rate generated by this method is two times higher compared to that of the ORPM-based animation generation method， so that the effect of the animation generated by the proposed method is more realistic and smooth.

Motion blurred image restoration algorithm based on multi-scale network

Haiyun WEI, Qianying ZHENG, Jinling YU

2022, 42(9):  2838-2844.  DOI: 10.11772/j.issn.1001-9081.2021081433

Asbtract ( )   HTML ( )   PDF (3756KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Non-uniform blind deblurring of dynamic scenes has always been a difficult problem in the field of image restoration. Aiming at the problem that the current blurred image restoration algorithms cannot solve the problem of diverse fuzzy sources well， an end-to-end motion blurred image restoration algorithm based on multi-scale network was proposed. In the proposed algorithm， the pruned residual blocks were used as the basic units， and the same asymmetric encoder-decoder network was used at each scale. In order to extract the features of the input image better， a residual module with attention mechanism was used in the coding side， and a spatial pyramid pooling layer was added. The recurrent unit between the encoding side and decoding side was able to obtain spatial information of the image， so that the image space continuity was able to be used to restore non-uniform motion blurred image. Test results show that the Peak Signal-to-Noise Ratio （PSNR） and Structural SIMilarity （SSIM） of the proposed algorithm are 33.69 dB and 0.953 7 respectively on GoPro dataset， and the blur image details can be recovered better， and the PSNR and SSIM of the proposed algorithm on Blur dataset are 31.47 dB and 0.904 7 respectively. Experimental results show that compared with scale-recurrent network and deep stacked hierarchical multi-patch network， the proposed algorithm achieves better blurred image restoration.

Image retrieval method based on deep residual network and iterative quantization hashing

Liefa LIAO, Zhiming LI, Saisai ZHANG

2022, 42(9):  2845-2852.  DOI: 10.11772/j.issn.1001-9081.2021071135

Asbtract ( )   HTML ( )   PDF (2416KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Focusing on the issue that the existing hashing image retrieval methods have weak expression ability， slow training speed， low retrieval precision， and difficulty in adapting to large-scale image retrieval， an image retrieval method based on Deep Residual Network and Iterative Quantitative Hashing （DRITQH） was proposed. Firstly， the deep residual network was used to perform multiple non-linear transformations on the image data to extract features of the image data and obtain high-dimensional feature vectors with semantic features. Then， Principal Component Analysis （PCA） was used to reduce the high-dimensional image features' dimensions. At the same time， to minimize the quantization error and obtain the best projection matrix， iterative quantization was used to binarize the generated feature vectors， the rotation matrix was updated and the data was mapped to the zero-center binary hypercube. Finally， the optimal binary hash code which was used to image retrieval in the Hamming space was obtained through performing the hash learning. Experimental results show that the retrieval precisions of DRITQH for four hash codes with different lengths on NUS-WIDE dataset are 0.789， 0.831， 0.838 and 0.846 respectively， which are 0.5， 3.8， 3.7 and 4.2 percentage points higher than those of Improved Deep Hashing Network （IDHN） respectively， and the average encoding time of the proposed method is 1 717 μs less than that of IDHN. DRITQH reduces the impact of quantization errors， improves training speed， and achieves higher retrieval performance in large-scale image retrieval.

Indoor dynamic scene localization and mapping based on target detection

Zhihong XI, Jiaxu WEN

2022, 42(9):  2853-2857.  DOI: 10.11772/j.issn.1001-9081.2021061077

Asbtract ( )   HTML ( )   PDF (1858KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem that dynamic objects in indoor scenes affect the accuracy of camera pose estimation seriously， a Simultaneous Localization And Mapping （SLAM） system for indoor dynamic scenes based on target detection was proposed. After the camera capturing an image， the YOLOv4 target detection network was used to detect dynamic objects in the environment and generate the mask area of the corresponding bounding box at first. Then， the ORB feature points in the image were extracted， and the feature points inside the mask area were removed. At the same time， the GMS （Grid-based Motion Statistics） algorithm was combined to further eliminate mismatches， and only the remaining static feature points were used to estimate the camera pose. Finally， the construction of a static dense point cloud map and an octomap filtering out dynamic objects was completed. Results of multiple comparison tests on TUM RGB-D public dataset show that compared to ORB-SLAM2 system， GCNv2_SLAM system and YOLOv4+ORB-SLAM2 system， the proposed system has the Absolute Trajectory Error （ATE） and Relative Pose Error （RPE） significantly reduced， indicating that this system can improve the accuracy of camera pose estimation in indoor dynamic environments significantly.

Residual attention deraining network based on convolutional long short-term memory

Zanxia QIANG, Xianfu BAO

2022, 42(9):  2858-2864.  DOI: 10.11772/j.issn.1001-9081.2021081379

Asbtract ( )   HTML ( )   PDF (3598KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Unmanned driving vehicles driving in rainy environment face the following problems： the images collected by the car on-board camera contain rain streak noise， which reduces the target detection accuracy and difficulty in identifying key targets of the unmanned driving system. In order to solve these problems， a residual attention deraining network based on convolutional long short-term memory was proposed. Firstly， the Convolutional Long Short-Term Memory （CLSTM） units were proposed to learn the distribution of different scales of rain streaks. Then， the residual channel attention mechanism was used to extract the rain streaks. Finally， the extracted rain streak information was subtracted from the rain image to obtain the restored background image. To determine the optimal network structure， the ablation experiments of each network module were carried out， and the structure with best rain removal effect was selected as the deraining network. Through the continuous optimization of network parameters， the proposed algorithm was tested on Rain100H， Rain100L and Real100 datasets， the results illustrate that the Peak Signal-to-Noise Ratio （PSNR） of the proposed algorithm reaches 29.1 dB， 33.1 dB and 32.4 dB respectively， and the Structural SIMilarity （SSIM） of the algorithm reaches 0.89， 0.94 and 0.93 respectively. Experimental results show that through the additional supervision of the Generative Adversarial Network （GAN） discriminator， the proposed algorithm achieves an visible rain streak removal effect and enhances the environmental perception ability of unmanned driving system under complex rainfall condition.

Depth estimation model of single haze image based on conditional generative adversarial network

Wentao ZHANG, Yuanyu WANG, Saize LI

2022, 42(9):  2865-2875.  DOI: 10.11772/j.issn.1001-9081.2021081386

Asbtract ( )   HTML ( )   PDF (7354KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To address the degradation problem of traditional depth estimation models caused by image quality degradation in haze environment， a model based on Conditional Generative Adversarial Network （CGAN） was proposed to estimate the depth of single haze image by fusing dual attention mechanism. Firstly， for the network structure of the generator of the model， the DenseUnet structure fused with dual attention mechanism was proposed. The dense blocks were used as basic blocks in the encoding and decoding processes of U-net. Dense and jump connections were used to enhance information flow， as well as extract the underlying structural features and high-level depth information of the direct transmission rate map. Then， the global dependencies of spatial features and channel features were adaptively adjusted by the dual attention module. At the same time， a new structure-preserving loss function was proposed by combining the least absolute value function， perceptual loss， gradient loss， and adversarial loss. Finally， using the direct transmission rate map of the haze image as a condition of CGAN， the depth map of the haze image was estimated through the adversarial learning of the generator and the discriminator. Training and testing were performed on the indoor dataset NYU Depth v2 and the outdoor dataset DIODE. Experimental results show that the proposed model has a finer geometric structure and richer local details. Compared with the fully convolutional residual network， on NYU Depth v2， the proposed model has the Logarithmic Mean Error （LME） and Root Mean Square Error （RMSE） error reduced by 7% and 10%， respectively. Compared with the deep ordinal regression network， on DIODE， the proposed model has the accuracy with threshold less than 1.25 increased by 7.6%. It can be seen that the proposed model improves the estimation accuracy and generalization ability of depth estimation under the interference of haze.

Traffic sign recognition model in haze weather based on YOLOv5

Jinghan YIN, Shaojun QU, Zekai YAO, Xuanye HU, Xiaoyu QIN, Pujing HUA

2022, 42(9):  2876-2884.  DOI: 10.11772/j.issn.1001-9081.2021071305

Asbtract ( )   HTML ( )   PDF (3770KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem of poor recognition precision and serious missed detection of small traffic signs in bad weather such as haze， rain and snow， a traffic sign recognition model in haze weather based on YOLOv5 （You Only Look Once version 5） was proposed. Firstly， the structure of YOLOv5 was optimized. By using contrary thinking， the problem of small object recognition difficulty was solved by reducing the depth of feature pyramid and limiting the maximum down sampling multiple. By adjusting the depth of residual module， the repeated overlapping of background features was suppressed. Secondly， the mechanisms such as data augmentation， K-means anchor and Global Non-Maximum Suppression （GNMS） were introduced into the model. Finally， the detection ability of the improved YOLOv5 facing the bad weather was verified on the Chinese traffic sign dataset TT100K， and the study on traffic sign recognition in the haze weather with the most obvious precision decline was focused on. Experimental results show that the F1-score， mean Average Precision @0.5 （mAP@0.5）， mean Average Precision @0.5：0.95 （mAP@0.5：0.95） of the improved YOLOv5 model reach 0.921 50， 95.3% and 75.2%， respectively. The proposed model can maintain high-precision recognition of traffic sign in bad weather， and has Frames Per Second （FPS） up to 50， meeting the requirement of real-time detection.

Multi-UAV real-time tracking algorithm based on improved PP-YOLO and Deep-SORT

Jun MA, Zhen YAO, Cuifeng XU, Shouhong CHEN

2022, 42(9):  2885-2892.  DOI: 10.11772/j.issn.1001-9081.2021071146

Asbtract ( )   HTML ( )   PDF (2914KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The target size of the Unmanned Aerial Vehicle （UAV） is small， and the characteristics among multiple UAVs are not obvious. At the same time， the interference of birds and flying insects brings a huge challenge to the accurate detection and stable tracking of the UAV targets. Aiming at the problem of poor detection performance and unstable tracking of small target UAVs by using traditional target detection algorithms， a real-time tracking algorithm for multiple UAVs based on improved PaddlePaddle-YOLO （PP-YOLO） and Simple Online and Realtime Tracking with a Deep association metric （Deep-SORT） was proposed. Firstly， the squeeze-excitation module was integrated into PP-YOLO detection algorithm to achieve feature extraction and detection of UAV targets. Secondly， the Mish activation function was introduced into ResNet50-vd structure to solve the problem of vanishing gradient in the back propagation process and further improve the detection precision. Thirdly， Deep-SORT algorithm was used to track UAV targets in real time， and the backbone network that extracts appearance features was replaced with ResNet50， thereby improving the original network’s weak perceptual ability of small appearances. Finally， the loss function Margin Loss was introduced， which not only improved the class separability， but also strengthened the tightness within the class and the difference between classes. Experimental results show that the detection mean Average Precision （mAP） of the proposed algorithm is increased by 2.27 percentage points compared to that of the original PP-YOLO algorithm， and the tracking accuracy of the proposed algorithm is increased by 4.5 percentage points compared to that of the original Deep-SORT algorithm. The proposed algorithm has a tracking accuracy of 91.6%， can track multiple UAV targets within 600 m in real time， and effectively solves the problem of "frame loss" in the tracking process.

Joint detection and tracking algorithm of target in aerial refueling scenes

Yi ZHANG, Yongrong SUN, Kedong ZHAO, Hua LI, Qinghua ZENG

2022, 42(9):  2893-2899.  DOI: 10.11772/j.issn.1001-9081.2021071286

Asbtract ( )   HTML ( )   PDF (4223KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Focusing on the target tracking problem in the docking stage of autonomous aerial refueling， a joint detection and tracking algorithm of target in aerial refueling scenes was proposed. In the algorithm， CenterTrack network with integrated detection and tracking was adopted to track the drogue. In view of the large computational cost and long training time， this network was improved from two aspects： model design and network optimization. Firstly， dilated convolution group was introduced into the tracker to make the network weight lighter without changing the size of the receptive field. At the same time， the convolutional layer of the output part was replaced with depthwise separable convolutional layer to reduce the network parameters and computational cost. Then， the network was further optimized to make it converge to a stable state faster by combining Stochastic Gradient Descent （SGD） method with Adaptive moment estimation （Adam） algorithm. Finally， videos of real-world aerial refueling scenes and simulations on the ground were made into dataset with the corresponding format for experimental verification. The training and testing were carried out on the self-built drogue dataset and MOT17 （Multiple Object Tracking 17） public dataset respectively， and the effectiveness of the proposed algorithm was verified. Compared to the original CenterTrack network， the improved network Tiny-CenterTrack reduces training time by about 48.6% and improves the real-time performance by 8.8%. Experimental results show that the improved network can effectively save the computing resources and improve the real-time performance to a certain extent without the loss of network performance.

Lightweight network for rebar detection with attention mechanism

Yaoshun LI, Lizhi LIU

2022, 42(9):  2900-2908.  DOI: 10.11772/j.issn.1001-9081.2021071136

Asbtract ( )   HTML ( )   PDF (4612KB) ( )  

Figures and Tables | References | Related Articles | Metrics

There are limited memory and computing power of the equipment in smart construction sites， making it very difficult to detect rebar in real time through object detection on the on-site equipment. The slow speed of rebar detection and the high cost of model deployment of this equipment also bring great challenges. In order to solve the problems， RebarNet， a lightweight network for rebar detection with attention mechanism was proposed on the basis of YOLOv3 （You Only Look Once version 3）. Firstly， the residual block was used as the basic unit of the network to construct a feature extraction structure to extract local and contextual information. Secondly， Channel Attention （CA） module and Spatial Attention （SA） module were added to the residual block to adjust the attention weight of the feature map and improve the ability of the network to extract features. Thirdly， the feature pyramid fusion module was used to increase the receptive field of the network and optimize the extraction effect of the medium-sized rebar images. Finally， the feature map of 52×52 channel was output for post-processing and rebar detection after 8 times downsampling. Experimental results show that the parameter amount of the proposed network is only 5% of that of Darknet53 network， and mAP （mean Average Precision） of the proposed network achieves 92.7% at the speed of 106.8 FPS （Frames Per Second） on the rebar test dataset. Compared with the existing 8 object detection networks including EfficientDet （Scalable and Efficient Object Detection）， SSD （Single Shot MultiBox Detector）， CenterNet， RetinaNet， Faster RCNN （Faster Region-CNN）， YOLOv3， YOLOv4 and YOLOv5m （YOLOv5 medium）， RebarNet has a shorter training time （24.5 seconds）， the lowest memory usage （1 956 MB）， and the smallest model weight file （13 MB）. Compared with the current best-performing YOLOv5m network， RebarNet has the mAP slightly lower by 0.4 percentage points with the detection speed increased by 48 FPS， which is 1.8 times of that of YOLOv5m network. The above indicates that the proposed network helps to complete the task of high-efficiency and accurate rebar detection in smart construction sites.

Image detection algorithm of cerebral arterial stenosis by improved Libra region-convolutional neural network

Hanqing LIU, Xiaodong KANG, Fuqing ZHANG, Xiuyuan ZHAO, Jingyi YANG, Xiaotian WANG, Mengfan LI

2022, 42(9):  2909-2916.  DOI: 10.11772/j.issn.1001-9081.2021071206

Asbtract ( )   HTML ( )   PDF (5263KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In view of the problems of vascular pleomorphism on transverse sections and sampling imbalance in the process of detection， an improved Libra Region-Convolutional Neural Network （R-CNN） cerebral arterial stenosis detection algorithm was proposed to detect internal carotid artery and vertebral artery stenosis in Computed Tomography Angiography （CTA） images. Firstly， ResNet50 was used as the backbone network in Libra R-CNN， Deformable Convolutional Network （DCN） was introduced into the 3， 4， 5 stages of backbone network， and the offsets were learnt to extract the morphological features of blood vessels on different transverse sections. Secondly， the feature maps extracted from the backbone network were input into Balanced Feature Pyramid （BFP） with the Non-local Neural Network （Non-local NN） introduced for deeper feature fusion. Finally， the fused feature maps were input to the cascade detector， and the final detection result was optimized by increasing the Intersection-over-Union （IoU） threshold. Experimental results show that compared with Libra R-CNN algorithm， the improved Libra R-CNN detection algorithm increases 4.3， 1.3， 6.9 and 4.0 percentage points respectively in AP， AP₅₀， AP₇₅ and AP_S， respectively_{on the cerebral artery CTA dataset； on the public CT dataset of colon polyps， the improved Libra R-CNN detection algorithm has the AP， AP50， AP75 and APS increased by 6.6， 3.6， 13.0 and 6.4 percentage points， respectively. By adding DCN， Non-local NN and cascade detector to the backbone network of Libra R-CNN algorithm， the features are further fused to learn the semantic information of cerebral artery structure and make the results of narrow area detection more accurate， and the improved algorithm has the ability of generalization in different detection tasks.}

Automatic detection algorithm for attention deficit/hyperactivity disorder based on speech pause and flatness

Guozhong LI, Ya CUI, Yixin EMU, Ling HE, Yuanyuan LI, Xi XIONG

2022, 42(9):  2917-2925.  DOI: 10.11772/j.issn.1001-9081.2021071213

Asbtract ( )   HTML ( )   PDF (1994KB) ( )  

Figures and Tables | References | Related Articles | Metrics

The clinicians diagnose Attention Deficit/Hyperactivity Disorder （ADHD） mainly based on on their subjective assessment， which lacks objective criteria to assist. To solve this problem， an automatic detection algorithm for ADHD based on speech pause and flatness was proposed. Firstly， the Frequency band Difference Energy Entropy Product （FDEEP） parameter was used to automatically locate the segment with voice from the speech and extract the speech pause features. Then， Transform Average Amplitude Squared Difference （TAASD） parameter was presented to calculate the voice multi-frequency and extract the flatness features. Finally， fusion features and the Support Vector Machine （SVM） classifier were combined to realize the automatic recognition of ADHD. The speech samples of the experiment were collected from 17 normal control children and 37 children with ADHD. Experimental results show that the proposed algorithm can effectively discriminate the normal children and children with ADHD， with an accuracy of 91.38%.

Frontier and comprehensive applications

Hybrid adaptive large neighborhood search algorithm for solving time-dependent vehicle routing problem in cold chain logistics

Zhihao XIAO, Zhihua HU, Lin ZHU

2022, 42(9):  2926-2935.  DOI: 10.11772/j.issn.1001-9081.2021071361

Asbtract ( )   HTML ( )   PDF (1487KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems of premature convergence and easily falling into local optimum in the adaptive large neighborhood search algorithms with single mechanism， a hybrid adaptive large neighborhood search algorithm was proposed to solve Time-Dependent Vehicle Routing Problem （TDVRP） in cold chain logistics. Firstly， the time-varying vehicle speed was described according to the continuous driving time dependent function， the real-time fuel consumption was evaluated by using the comprehensive fuel consumption model， and a routing optimization model with the goal of minimizing the total cost was established. Then， according to the NP （Non-deterministic Polynomial）-hard property and time-dependent characteristics of the problem， a variety of large neighborhood search operators for destroying and repairing solutions were designed， and the destroy-repair large neighborhood search operators were integrated into Artificial Bee Colony （ABC） algorithm to improve the global search ability of the algorithm. Simulation results show that compared with Adaptive Variable Neighborhood Search Elite Ant Colony （AVNS_EAC） algorithm， Adaptive Large Neighborhood Search Elite Ant Colony （ALNS_EAC） algorithm， Adaptive Large Neighborhood Search Elite Genetic （ALNS_EG） algorithm and Adaptive Large Neighborhood Search Simulated Annealing （ALNS_SA） algorithm， the proposed Adaptive Large Neighborhood Search Artificial Bee Colony （ALNS_ABC） algorithm has the optimal fitness values increased by 46.3%， 5.3%， 36.8% and 6% respectively and averagely on multiple test data groups. It can be seen that this algorithm has higher computational performance and stronger stability， and can provide a more reasonable decision-making basis for cold chain logistics enterprises to take into account economic and environmental benefits at the same time.

Car-following model of intelligent connected vehicles based on time-delayed velocity difference and velocity limit

Kaiwang ZHANG, Fei HUI, Guoxiang ZHANG, Qi SHI, Zhizhong LIU

2022, 42(9):  2936-2942.  DOI: 10.11772/j.issn.1001-9081.2021081425

Asbtract ( )   HTML ( )   PDF (2663KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Focusing on the problems of disturbed car-following behavior and instability of traffic flow caused by the uncertainty of the driver’s acquisition of road velocity limit and time delay information， a car-following model TD-VDVL （Time-Delayed Velocity Difference and Velocity limit） was proposed with the consideration of the time-delayed velocity difference and the velocity limit information in the Internet of Vehicles （IoV） environment. Firstly， the speed change caused by time delay and road velocity limit information were introduced to improve the Full Velocity Difference （FVD） model. Then， the linear spectrum wave perturbation method was used to derive the traffic flow stability judgment basis of TD-VDVL model， and the influence of each parameter in the model on the stability of the system was analyzed separately. Finally， the numerical simulation experiments and comparative analysis were carried out using Matlab. In the simulation experiments， straight roads and circular roads were selected， and slight disturbance was imposed on the fleet during driving. When conditions were the same， TD-VDVL model had the smallest velocity fluctuation rate and the fluctuation of fleet headway compared to the Optimal Velocity （OV） and FVD models. Especially when the sensitivity coefficient of the velocity limit information was 0.3， and the sensitivity coefficient of the time-delayed speed difference was 0.3， the proposed model had the average fluctuation rate of the fleet velocity reached 2.35% at time of 500 s， and the peak and valley difference of fleet headway of only 0.019 4 m. Experimental results show that TD-VDVL model has a better stable area after introducing time-delayed velocity difference and velocity limit information， and can significantly enhance the ability of car-following fleet to absorb disturbance.

Optimization model of inventory system under stochastic disturbance based on active disturbance rejection control

Chuan ZHAO, Luyao LI, Haoxiong YANG, Min ZUO

2022, 42(9):  2943-2951.  DOI: 10.11772/j.issn.1001-9081.2021071303

Asbtract ( )   HTML ( )   PDF (3630KB) ( )  

Figures and Tables | References | Related Articles | Metrics

To solve the problem of stockout， increasing inventory level and the fluctuation of order quantity caused by stochastic disturbance， an optimization model of inventory system under stochastic disturbance based on Active Disturbance Rejection Control （ADRC） was proposed. Firstly， according to the operational management logic behind the purchase-sale-storage product and information flows， the transfer function of the inventory system was obtained and transformed to a second-order state space standard form by the Laplace transform. Secondly， an optimization model of inventory system under stochastic disturbance based on ADRC including the tracking differentiator， the extended state observer and the nonlinear state error feedback control law was designed to control and compensate the adverse effects on the inventory system caused by stochastic disturbance under the premise of ensuing system stability. Finally， simulations were carried out by using data collected from the industry to verify the effectiveness of the optimization model on optimization of the inventory system under stochastic disturbance. Simulation results show that compared to the inventory feedback control model without ADRC， the optimization model of inventory system under stochastic disturbance based on ADRC has the residual inventory reduced by 40%， the average order quantity reduced by 47.4%， the order fluctuation decreased by 39.3%， and the stockout of enterprise inventory system caused by stochastic disturbance greatly improved. It can be seen that the optimization model of inventory system under stochastic disturbance based on ADRC can guide enterprises to make a reasonable ordering decision， decrease the inventory level， improve the stability of inventory system dynamically， and provide the scientific theoretical reference and countermeasures for the actual operations of enterprises.

Hybrid bird swarm algorithm for solving permutation flowshop scheduling problem

Hongchao YAN, Wei TANG, Bin YAO

2022, 42(9):  2952-2959.  DOI: 10.11772/j.issn.1001-9081.2021091650

Asbtract ( )   HTML ( )   PDF (2360KB) ( )  

Figures and Tables | References | Related Articles | Metrics

A Hybrid Bird Swarm Algorithm （HBSA） was proposed to minimize the makespan more efficiently for Permutation Flowshop Scheduling Problem （PFSP）. Firstly， to improve the quality and diversity of initial population， a new population initialization method was put forward by combining a NEH （Nawaz-Enscore-Ham） based heuristic algorithm and chaotic mapping. Secondly， to deal with the discrete scheduling problem by the algorithm， the Largest Ranked Value （LRV） rule was adopted to convert continuous position values to discrete job permutation. Finally， to enhance the ability of the algorithm to explore the solution space， local search methods for the individual best job permutation and population best job permutation were proposed on the basis of the ideas of Variable Neighborhood Search （VNS） and Iterative Greedy （IG） algorithms respectively. The proposed algorithm was simulated and tested on the widely used benchmark test set Rec and compared with Hybrid Differential Evolution algorithm proposed by Liu et al （L-HDE） algorithm， Hybrid Symbiotic Organisms Search （HSOS） algorithm， Discrete Wolf Pack Algorithm （DWPA） and Multi-Class Teaching-Learning-Based Optimization （MCTLBO） algorithm， which are the effective meta-heuristic algorithms for PFSP. The results show that the average values of Best Relative Error （BRE） and Average Relative Error （ARE） achieved by HBSA are at least 73.3% and 76.8% lower than those of the above four algorithms， thus proving that HBSA has stronger search ability and better stability. It is worth mentioning that， for Rec25 and Rec27 test instances， only HBSA achieves the currently known optimal solutions， which further proves its superiority.

Remaining useful life prediction method of aero-engine based on optimized hybrid model

Yuefeng LIU, Xiaoyan ZHANG, Wei GUO, Haodong BIAN, Yingjie HE

2022, 42(9):  2960-2968.  DOI: 10.11772/j.issn.1001-9081.2021071343

Asbtract ( )   HTML ( )   PDF (2754KB) ( )  

Figures and Tables | References | Related Articles | Metrics

In the Remaining Useful Life （RUL） prediction methods of aero-engine， the data at different time steps are not weighted simultaneously， including the original data and the extracted features， which leads to the problem of low accuracy of RUL prediction.Therefore， an RUL prediction method based on optimized hybrid model was proposed. Firstly， three different paths were chosen to extract features. 1） The mean value and trend coefficient of the original data were input into the fully connected network. 2） The original data were input into Bidirectional Long Short-Term Memory （Bi-LSTM） network， and the attention mechanism was used to process the obtained features. 3） The attention mechanism was used to process the original data， and the weighted features were input into Convolutional Neural Network （CNN） and Bi-LSTM network. Then， the idea of fusing multi-path features for prediction was adopted， the above-mentioned extracted features were fused and input into the fully connected network to obtain the RUL prediction result. Finally， the Company-Modular Aero-Propulsion System Simulation （C-MAPSS） datasets were used to verify the effectiveness of the method. Experimental results show that the proposed method performs well on all the four datasets. Taking FD001 dataset as an example， the Root Mean Square Error （RMSE） of the proposed method is reduced by 9.01% compared to that of Bi-LSTM network.

Noctiluca scintillans red tide extraction method from UAV images based on deep learning

Jinghu LI, Qianguo XING, Xiangyang ZHENG, Lin LI, Lili WANG

2022, 42(9):  2969-2974.  DOI: 10.11772/j.issn.1001-9081.2021071197

Asbtract ( )   HTML ( )   PDF (3025KB) ( )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problems of low accuracy and poor real-time performance of Noctiluca scintillans red tide extraction in the field of satellite remote sensing， a Noctiluca scintillans red tide extraction method from Unmanned Aerial Vehicle （UAV） images based on deep learning was proposed. Firstly， the high-resolution RGB （Red-Green-Blue） videos collected by UAV were used as the monitoring data， the backbone network was modified to VGG-16 （Visual Geometry Group-16） and the spatial dropout strategy was introduced on the basis of the original UNet++ network to enhance the feature extraction ability and prevent the overfitting respectively. Then， the VGG-16 network pre-trained by using ImageNet dataset was applied to perform transfer learning to increase the network convergence speed. Finally， in order to evaluate the performance of the proposed method， experiments were conducted on the self-built red tide dataset Redtide-DB. The Overall Accuracy （OA）， F1 score， and Kappa of the Noctiluca scintillans red tide extraction of the proposed method are up to 94.63%， 0.955 2， 0.949 6 respectively， which are better than those of three traditional machine learning methods — K-Nearest Neighbors （KNN）， Support Vector Machine （SVM） and Random Forest （RF） as well as three typical semantic segmentation networks （PSPNet （Pyramid Scene Parsing Network）， SegNet and U-Net）. Meanwhile， the red tide images of different shooting equipment and shooting environments were used to test the generalization ability of the proposed method， and the corresponding OA， F1 score and Kappa are 97.41%， 0.965 9 and 0.938 2， respectively， proving that the proposed method has a certain generalization ability. Experimental results show that the proposed method can realize the automatic accurate Noctiluca scintillans red tide extraction in complex environments， and provides a reference for Noctiluca scintillans red tide monitoring and research work.