Search

In order to solve the problems of the current art circulation information is not transparent and the centralized trading authority is asymmetric, this paper proposed a art blockchain certificate traceability model based on three chains. First, in view of the large amount of data in a single piece of artwork information, this paper proposed a blockchain-based symmetric searchable encryption algorithm, and used the interplanetary file system to securely store artwork circulation records on-chain and off-chain. Secondly, for the characteristics of the increasing effective data of artwork information, this paper proposed a blockchain-based inverted fast query and traceability algorithm to help data users realize fast traceability and positioning. At the same time, this paper proposed for the first time that the false information alliance chain stores false art circulation records to realize the processing of false information. Finally, experiments verify the feasibility of the model for big data security storage and fast query traceability. Experiments show that the model is feasible and the query method is efficient. [ABSTRACT FROM AUTHOR]

This paper proposed an electronic voting scheme based on one-time ring signature and blockchain to solve the problems of the current e-voting system in terms of anonymity, transparency and open vote-counting. This paper introduced one-time ring signature, blockchain and stealth address technology. The one-time ring signature technology solved the problem of voter anonymity, repeated voting and public verification. The blockchain technology could ensure the openness and transparency of the e-voting system. The one-time ring signature technology could ensure no one or institution could obtain the voting intermediate result before the voting ends. In order to improve the efficiency of vote-counting,it used the method of key sharing to improve the vote-counting algorithm. Through security and performance analysis, the scheme conforms to the basic security standards of an electronic voting system. [ABSTRACT FROM AUTHOR]

Regular path query is a technique of using regular expressions to query graph data, usually uses finite state automata to achieve query matching. The matching result of the existing regular path query method is a sequence of vertex pairs, and the structure of the graph is not fully preserved. In order to solve this problem, this paper proposed a method of structured regular path query for graph data. By structuring constraints between different sequences, the query results could be transformed from paths to subgraphs. For this purpose, the method firstly defined a structured regular path query language, and then designed a structured query parsing and a matching algorithm based on this structure. Experimental results on simulated and real datasets verify the influence of network size on query speed. This paper compared the proposed method with the control group. Experimental results show that the proposed method can realize structured query under the premise of satisfying regular expression constraints. [ABSTRACT FROM AUTHOR]

In reinforcement learning, the policy gradient algorithm often needs to model the continuous-time process as a discrete-time process through sampling. To model the problem more accurately, it improves the sampling frequency. However, the excessive sampling frequency may reduce the training efficiency. To solve this problem, this paper proposed action stable updating algorithm. This method calculated the probability of action repetition using the change of the output of the policy function, and randomly repeated or changed the action based on this probability. This paper theoretically analyzed the performance of this method. This paper evaluated the performance of this method in nine different environments and compared it with the existing methods. This method surpassed existing methods in six of these environments. The experimental results show that this method can improve the training efficiency of the policy gradient algorithm in continuous-time problems. [ABSTRACT FROM AUTHOR]

Aiming at the shortcomings of whale optimization algorithm (WOA) when solving problems such as high dimension, multi-peak and non-origin of optimal value, such as low convergence accuracy and easy to be captured by local optimum, this paper proposed a whale optimization algorithm based on multi-dimensional variation learning and distributed optimization (MLDOWOA) . Firstly, this algorithm introduced adaptive weights and dominant individual interference to dynamically adjust the direction of individual spiral surround, which improved the global search ability and convergence accuracy of the algorithm. Then, in order to further expand the search range of the algorithm, the improved algorithm used a multi-dimensional variation learning mechanism to adaptively plan the direction of population variation. Finally, it put forward the distributed optimization mechanism to coordinate the step size of search, so as to help the population break through the limitation of search stagnation in the middle and late stage. The test results on 8 high dimensional benchmark functions and 4 fixed dimensional benchmark functions show that, compared with the basic WOA, SSA and the improved WOA algorithms ACWOA, AWOA, MSIWOA, ADWOA, the proposed algorithm has significant advantages in convergence accuracy and ability to cope with high dimensional functions. This paper used the algorithm to tune the parameters of the FOPID controller, and compared the control results with the research results of this engineering problem in recent years, which prove the excellent performance of the proposed algorithm in FOPID parameter tuning problems. [ABSTRACT FROM AUTHOR]

Current bitmap collaboration models suffer from issues such as large volume of data transmission, poor flexibility, overlapping conflicts and storage pressure. To solve these problems, this paper proposes a collaborative editing model for complex bitmaps. Collaborative operation that sends only the coordinates and color change, featuring low data volume, multi-dimensional granularity, and overlap able. Adds a four-dimensional bitmap collaborative architecture with spatiotemporal relationship, and only saves the edited pixel data, which can save more storage space; A replica state control algorithm based on rollback that ensures operations are executed on correct replica. An algorithm that can dynamically generate branch versions according to the current state and policy configuration, solving the problem of too many branch versions. Finally, inferring model correctness based on the CCI model, develop a prototype collaborative editor “Co-Pixel”, execute experiments based on randomly generated operation sets, simulation of actual editing behavior against Adobe Lightroom. Results show that this model can avoid conflicts between independent operations to a greater extent, can reduce the algorithmic overhead in the consistency maintenance process, can flexibly adjust the intent retention policy, and can generate a version of the result that reduces the user’s intent maintenance effort. The model in this paper provides a good solution for collaborative bitmap editing. [ABSTRACT FROM AUTHOR]

After the outbreak of the epidemic, the distribution of living supplies in residential areas under closed management become one of the focal issues to be solved urgently, and this problem can be abstracted as the the location problem of distribution points for living supplies during the epidemic, whose essence is the NP-Hard problem in combinational optimization. This paper studied the exact algorithm of the location problem of distribution points for living supplies during the lockdown period. Firstly, the paper researched some mathematical properties and corresponding proof of the problem, and these properties could be used to reduce the size of the given problem, thereby reducing the difficulty of the problem solving. Then the paper designed allocation sub-algorithm, upper and lower bound sub-algorithms and the reduced order sub-algorithm. Based on these sub-algorithms, the paper proposed a backtracking algorithm with reduction which can reduce the size of the problem and get the optimal solution. Finally,the paper analyzed and solved several examples to further illustrate the principle and execution process of the proposed algorithm. The results show that the proposed algorithm can reduce the difficulty of solving the problem by reducing the size of the problem. [ABSTRACT FROM AUTHOR]

In order to achieve a fair distribution of benefits in federated learning, an indicator that quantifies the contribution of each data provider to the global model is critical. Aiming at the problems of privacy leakage, opacity and dependence on central server in existing contribution evaluation schemes, this paper proposed a transparent federated learning contribution evaluation scheme based on blockchain. Firstly, this paper proposed an improved Paillier secure aggregation algorithm, which avoided the inference of user local data in the model aggregation stage by joint decryption. Secondly, this paper proposed a method to approximate the contribution based on the gradient of user cumulative submission, which solved the problem of privacy leakage in the existing contribution evaluation scheme. In addition, it integrated the evaluation of the contribution into the consensus process of the blockchain, making the evaluation results auditable. Finally, experiments based on MNIST dataset show that the proposed method can effectively evaluate the contribution. [ABSTRACT FROM AUTHOR]

In order to address the shortcomings of the facial landmarks detection models, which cannot model the relations between long-distance landmarks, this paper proposed a parallel multi-branch architecture combining with convolution and Transformer for facial landmarks tasks, called MCTN, it utilized the dynamic attention mechanism to model the long-distance relations between facial landmarks. The multi-branch parallel structure designing allowed MCTN to include shared weights, global information fusion and other merits. What’s more, this paper proposed the novel Transformer structure, Deformer, which could make the MCTN focused attention weights faster on sparse and meaningful locations and solved the problem of slow convergence of Transformer. MCTN reached 4.33%,3.12% and 3.15% normalized average error respectively on the WFLW,300W and COFW datasets, the results show that MCTN utilizes Transformer with CNN multi-branch parallel structure and Deformer structure to dramatically outperform other facial landmarks localization algorithms based on convolution network. [ABSTRACT FROM AUTHOR]

A variety of factors will affect the combination and distribution of the molds used in the production of prefabricated components on the mold table, and will produce many combination and distribution results. In order to solve such problems and quickly find the best distribution results of the molds on the mold table, this paper proposed a mold combination method based on the combination of backtracking method and improved Eclat algorithm. Firstly, it built a binary tree model based on the mold combination relationship in the historical data, searched the binary tree through the backtracking method, and obtained all theoretically feasible mold combination results. Then it pre-placed these combination results by using the improved BL positioning algorithm on the mold table with a fixed area, and screened out the mold combination results that met the actual placement conditions. It used Eclat algorithm to obtain the optimal combination set of mold adaptation from the selected combination results, determined the association rules between different types of molds,and used the rules to guide the allocation process of molds on the mold table in the new component production task. At the same time, this paper improved the Eclat algorithm by introducing the partition idea and the priori constraint idea based on probability, so as to improve its operation efficiency when dealing with large-scale data. Finally, through the simulation test of the case data in the prefabricated component manufacturing enterprise, it verified the effectiveness of the method to quickly solve the problem of mold table combination allocation. The results show that the mold combination allocation based on backtracking and DEclat algorithm method has a good effect on the utilization of the mold area, the reduction of the mold changing time and the reduction of the total completion time. [ABSTRACT FROM AUTHOR]

To address the shortcomings of the object detection model in the training process, such as the allocation of positive and negative samples without considering the aspect ratio of the ground-truth box and the poor adaptability to different distributions of objects, this paper proposed the ratio-prior and loss-aware assignment（RLA） algorithm. RLA didn’t change the structure of the original detection model, firstly it selected an equal proportion of the central sampling area based on the aspect ratio of the ground-truth box, then calculated the integrated loss of the anchor points, considered the actual distribution of objects within the ground-truth box, and finally distinguished between positive and negative samples by means of a dynamic loss threshold. The algorithm solved the problems of poor adaptability and difficulty in selecting the best positive samples based on IoU allocation, and the sample allocation for eccentric objects and objects with different aspect ratios was more reasonable. Compared with existing sample allocation algorithms, the algorithm outperforms the MS COCO dataset by 1. 66% over the AP of the baseline FCOS, and 0. 76% and 0. 24% over the AP of the ATSS algorithm and the PAA algorithm respectively when the model structure is the same, demonstrating the effectiveness of the RLA algorithm. [ABSTRACT FROM AUTHOR]

In the recommendation system, the user’s ratings are often affected by the rating context, that is, the user’s previous ratings of some items will affect the objectivity of his rating of the current item. Sparse linear method treats user ratings affected by context as the same as other ratings when calculating item similarity. However, this partial ratings cannot objectively reflect the similarity between items. To solve the above problems, this paper proposed a recommendation algorithm combining rating context and item similarity based on sparse linear method. It divided the algorithm into three stages.The first stage used weighted ratings to calculate the item’s nearest neighbor for feature selection.In the second stage, it used the rating error weight to reduce the fitting of the ratings affected by the context of the algorithm model, and trained the item similarity matrix. In the third stage, it predicted the ratings according to the user’s ratings and the item similarity, and finally sorted the predicted ratings to complete the item recommendation. Experiments were conducted on four datasets of MovieLens, it used mean average precision (MAP), mean reciprocal rank (MRR) and normalized discounted cumulative gain (NDCG) to evaluate the effectiveness of the algorithm. The experimental results show that the fusion rating context will further improve the accuracy of item similarity and thus improve the performance of recommendation. [ABSTRACT FROM AUTHOR]

Road traffic accidents occur frequently and seriously in the accident-prone spots. In order to improve the safety and efficiency of road traffic, it is necessary to find the location of accident-prone points. The existing density clustering algorithm needed to set the number of center points and was easy to expand the clustering range when identifying traffic accident-prone points, this paper proposed the limit cluster expansion and adaptive search clustering by fast search and find of density peaks (LA-CFDP) algorithm to solve these problems. LA-CFDP algorithm automatically determined the number of center points by increasing the restriction condition of center points, and introduced the parameter expansion factor to limit the cluster expansion range, so as to improve the adaptability and accuracy of the algorithm for accident-prone point identification. The case analysis on the 2019 traffic accident data set of four cities in United Kingdom shows that the Sihouette coefficient of the clustering results of LA-CFDP algorithm reaches 0.72～0.92, and the Davies-Bouldin index (DBI) are all reduced to below 0.37. The clustering results accord with the definition and characteristics of accident-prone spots, and can provide reliable basis for the management of accident-prone spots. [ABSTRACT FROM AUTHOR]

When the meta-heuristic algorithm solves the joint resource allocation problem of multi-beam satellites, the computational complexity increases and the algorithm is difficult to converge due to the time delay constraint and capacity constraint. This paper introduced a penalty mechanism in the objective function, and added a penalty value to the objective function of the invalid solution, so that the optimized solution adaptively satisfied these two constraints. Based on this, this paper proposed a joint resource allocation algorithm based on quantum-behaved particle swarm optimization. Simulation results show that the introduction of the penalty strategy solves the problem of difficulty in handling the delay constraint and capacity constraint when applying the meta-heuristic algorithm. The quantum-behaved particle swarm optimization algorithm with penalty mechanism outperforms the existing joint allocation algorithm in terms of allocation fairness index and total system capacity. [ABSTRACT FROM AUTHOR]

In order to solve the problems of low convergence accuracy and difficulty in jumping out of local optimum of traditional SSA, this paper proposed an improved salp swarm algorithm (ISSA) with multi-strategy fusion. Firstly, this paper proposed a new follower position update method integrating the convergence strategy of the midperpendicular algorithm to solve the shortcomings of the traditional SSA follower position update method. In order to improve the ability of SSA to jump out of the local optimum, this paper proposed a self perturbation strategy based on the convergence strategy of the midperpendicular algorithm. Secondly, by analyzing the shortcomings of the traditional SSA leader position update strategy, this paper proposed a new leader position update strategy, and for the fixed population order of SSA, this paper proposed a method to reconstruct the arrangement order of the salp population with the index of fitness to improve the performance of the algorithm. Finally, simulation experiments prove the effectiveness of the ISSA, and the experiments results show that ISSA solves the problems of low convergence accuracy and difficulty in jumping out of local optimum, and improves the convergence speed and stability of SSA. The superiority of ISSA is proved by comparison experiments with other improved SSA. [ABSTRACT FROM AUTHOR]

Although reinforcement learning shows good performance in game playing, system control and other fields, how to use a small number of samples to learn new tasks quickly is an urgent problem to be solved in reinforcement learning. At present, applying meta learning to reinforcement learning has been one of the most effective solutions, and the generated meta reinforcement learning has increasingly become a research hotspot in the field of reinforcement learning. To help researchers understand the field of meta reinforcement learning quickly, this paper sorted out the algorithms according to the literatures of meta reinforcement learning in recent years, summarized them into CNN-based meta-RL, context-based meta-RL, gradient-based meta-RL, hierarchical-based meta-RL and offline meta-RL and compared five types of algorithms. In addition, it briefly described the basic theories and challenges of meta reinforcement learning. Finally, this paper also discussed the future development of meta reinforcement learning based on the current research status. [ABSTRACT FROM AUTHOR]

This paper proposed an improved RRT（Rapidly-Explorative Random Tree） algorithm to solve the problem that the traditional target offset RRT was easy to fall into local optimum in the environment where there were many obstacles such as yarn racks, placing platforms and other packages during the package handling process. The collision detection model was established by geometric envelope method, and the distance coefficient g was added to keep a safe distance between the end of the manipulator and the obstacle; A random action selection strategy was proposed to select the expansion action before each expansion of the tree; A weakening target offset strategy was proposed to offset the extension direction according to the position of Xgoal relative to the Xnear and Xrand line; According to the vertical distance limit method and cubic spline interpolation, the initial path was optimized, and finally a smooth path was obtained. Through simulation comparison with RRT algorithm, M-RRT algorithm, RRT algorithm with added gravity coefficient and Improved RRT * algorithm, the maximum decrease in path length was 23.3%, the maximum decrease in time was 82.5%, and the distance from obstacles was always more than 50 mm. The results show that the algorithm improves the efficiency and safety of the manipulator handling. [ABSTRACT FROM AUTHOR]

The data accuracy and anti-interference of inertial measurement unit (IMU) three-axis Euler angle are affected by high frequency noise and strong instantaneous interference. To solve this problem, this paper proposes a dynamic error suppression method with low computational burden, good real-time performance and low cost, which is more suitable for embedded systems. In this method, an infinite impulse response-extended Kalman filter (IR-EKF) is introduced to the front end of the EKF algorithm. The data are preprocessed with a second-order low-pass Butterworth filter (BF) to help EKF suppress high frequency or strong interference. The IIR-EKF algorithm is implemented in the STM32H743 microcontroller. The experimental results show that the data variance is very big when the EKF acts alone, and great discrepancy occurs when strong interferences are encountered. When unscented Kalman filter (UKF)acts alone, the data variance is smaller than that of EKF. Although it does not depend on the initial noise parameters, it does not meet the requirements. After the addition of the second-order BF, the data variance is significantly reduced, the instantaneous error is greatly suppressed, and the stability and anti-interference ability of the system are enhanced. [ABSTRACT FROM AUTHOR]

In order to solve the problems of uneven resource utilization and long task completion time in complex task group scheduling, this paper constructed a complex task group resource scheduling model to minimize the mean square error of resource load and the task group completion time, and proposed a multi-objective optimization algorithm based on boundary range local search and NSGA-Ⅱ, called BRLSN. The algorithm used an effective coding method and cross mutation operator for iterative optimization, and constructed an elite retention strategy based on local search in boundary region to expand the search scope of the algorithm and preserved good individuals in the population. Experimental results show that compared with other multi-objective algorithms, the convergence and diversity of BRLSN are significantly improved. At the same time, the algorithm convergence speed is faster, the population quality is higher, and the result value of the final objective function is obviously optimized. [ABSTRACT FROM AUTHOR]

Most existing traditional congestion control algorithms are difficult to adapt to the highly dynamic link environment of wireless Ad hoc network. In order to solve the above problem, this paper proposed a method of improving the performance of congestion control based on deep reinforcement learning, Enhanced-CC. It conducted a preliminary detection of the congestion window by using the traditional congestion control algorithm. On this basis, the method used deep reinforcement technology to learn the real-time optimal congestion window range of the link. When the congestion window calculated by the traditional congestion control algorithm was too large or too small, the method adjusted the congestion window, so that the sending rate could match the highly dynamic link bandwidth, and the method could improve the transmission performance of the traditional congestion control algorithm. The experimental results show that Enhanced-CC can significantly improve the performance of traditional congestion control algorithms such as BBR, CUBIC, Westwood, Reno, and is superior to the performance of fully learning based congestion control algorithms such as PCC and PCC Vivace, and the combination of deep reinforcement learning and traditional congestion control algorithms such as Orca and DeepCC. [ABSTRACT FROM AUTHOR]

In order to fix the integer ambiguity quickly and accurately, this paper proposed a new ambiguity search algorithm based on measurement accuracy constraints to solve the problems of the LAMBDA algorithm, such as the wide search range of the integer ambiguity and the low search efficiency. When the least squares problem got optimally weighted, the algorithm took the carrier phase measurement accuracy as the constraint condition to test the fixed ambiguity solution on the basis of the stan-dard differential evolution （DE） algorithm. This algorithm solved the influence of different satellite altitude angles, and could achieve 99% success rate when solving 3D integer ambiguity. Compared with the MLAMBDA algorithm, the standard DE algorithm and the adaptive weighted differential evolution （AWDE） algorithm, the proposed algorithm further improves the efficiency and success rate of ambiguity resolution. [ABSTRACT FROM AUTHOR]

In order to solve the problem of low accuracy of 6D pose estimation for textured models under occlusion, this paper proposes an end-to-end 6D pose estimation algorithm based on local feature representation. Firstly, this paper proposes a Spatial and Coordinate Attention mechanism to obtain accurate localization information. A YOLOv5-CBE detection network forms by adding the attention mechanism to the backbone network and introducing a weighted Bidirectional Feature Pyramid Network in the detection layer. The Precision, Recall and mAP@0.5 of YOLOV5-CBE algorithm rise by 3.6%, 2.8% and 2.5% respectively, and the coordinate error of local feature center point decreases by 25% at most. Secondly, the YOLOv5-CBE network detects the local feature key points and calculates 6D pose of the model with 3 D Harris key points by Singular Value Decomposition, and the algorithm can guarantee 2D reprojection Accuracy and ADD Accuracy above 95% with 70% occlusion, which has a strong robustness. [ABSTRACT FROM AUTHOR]

Existing task offloading strategy usually makes offloading decision within one time slot without considering the internal relationship between multiple offload time slots, so they cannot be offloaded according to the actual needs of tasks. To solve this problem, this paper proposed a task secondary application offloading strategy based on deep Q network（DQN-TSAO）. Firstly, this paper introduced a three-layer of cloud-edge-end architecture that supports task secondary application offloading, and established priority model, delay model and energy consumption model for task offloading. Second, aiming at minimizing system energy consumption, the energy consumption optimization problem is transformed into a Markov decision process problem of maximum cumulative offloading reward. Finally, DQNTSAO algorithm can extract the task offload characteristics of each time slot, which enabled the task to obtain the optimal offloading decision of multiple time slots in the continuous interaction with the environment. Simulation results validated that DQN-TSAO algorithm can effectively reduce the total energy consumption of the system in a period of time. [ABSTRACT FROM AUTHOR]

At present, the research on combining deep learning with Visual SLAM(Simultaneous Localization And Mapping) to construct indoor 3 D pointcloud semantic map is in full swing. In order to solve the problems that traditional Visual SLAM networks have low accuracy and poor speed and lack semantic information, this paper proposed a novel RGB-D semantic segmentation network. The network uses the depth information which less affected by the light in indoor scenes to improve the accuracy of segmentation, and meanwhile designs the lightweight multi-scale residual module(MRAM) and Atrous Spatial Pyramid Pooling(ASPP) module to lightweight the segmentation network and improve the segmentation accuracy. Firstly, ORB-SLAM2 network screens the input image sequences to obtain keyframes. Then, the keyframes go into the proposed semantic segmentation network to get the 2 D semantic label, and then map the 2 D semantic information to 3 D pointcloud space. Finally, the method uses the Bayesian algorithm to update the 3 D map to obtain the globally consistent 3 D pointcloud semantic map. The experiments adopt NYUv2 dataset to verify the performance of semantic segmentation network, and adopt TUM dataset construct pointcloud semantic map. The results show that the performance and speed of the semantic segmentation network in this paper are better than the existing models, and the combination of semantic segmentation network with visual SLAM can meet the requirements of constructing 3 D dense semantic pointcloud map accurately and quickly. [ABSTRACT FROM AUTHOR]

To solve the problems of low convergence accuracy, slow convergence speed and easy to fall into local optimum of traditional particle swarm algorithm, this paper proposed an improved particle swarm algorithm with multistrategy fusion. Firstly, in order to accelerate the convergence speed of free particles, the improved algorithm used a method of updating the position of free particles based on the midperpendicular algorithm. secondly, the improved algorithm designed a strategy of generating exploding particles near the optimal particles to enhance the optimizationseeking accuracy and optimization-seeking speed of the algorithm, and the improved algorithm also designed a particle velocity updating strategy relying only on the global optimal particle position to accommodate the first two strategies. Finally, The algorithm also used the inertia weights and particle position update methods of the simplified particle swarm optimization algorithm based on probabilistic hierarchy. This paper designed a few comparison experiments with other five improved particle swarm algorithms, and the results show that the improved algorithm in this paper has a greater advantage and better performance whether dealing with low-dimensional problems or high-dimensional problems. [ABSTRACT FROM AUTHOR]

The homology analysis of diffusion results aims to identify which historical diffusion results were produced by the same set of diffusion sources. This paper uses Bernoulli Mixture Model to solve this problem: The diffusion results generated by the same group of sources correspond to the same Bernoulli component, and the parameters of the corresponding Bernoulli component reflects the arrival probability of the sources’ influence at each point. Then the log-likelihood function is constructed based on the model and the observed data. To determine the results of homology analysis, the Bernoulli parameters in the log-likelihood function are calculated through using the EM algorithm. A large number of experimental results verify that compared with the traditional clustering algorithm, the algorithm in this paper can get higher accuracy of the homologous clustering. That is to say, the algorithm in this paper can perform homology analysis on the propagation result data more efficiently and accurately. [ABSTRACT FROM AUTHOR]

Aiming at the problems that the Raft algorithm cannot resist the attacks of Byzantine nodes and the logs are easy to tamper with, this paper proposes an RB-Raft (Resist Byzantine-Raft) algorithm that resists Byzantine nodes. Firstly, this paper uses the method of hash chain to iteratively hash each log. At the same time, verification of the log through the dynamic verification mechanism, so that the malicious behavior of the leader node has a certain fault tolerance rate, which solves the problem of log forgery and verification. Secondly, this paper proposes a "Legacy" mechanism based on threshold encryption, which makes it legal for Candidate nodes to pull votes. This mechanism can prevent Byzantine nodes from randomly the attack of pulling votes to replace leader nodes, and solves the problem that Byzantine nodes affect the system consistency. The experimental results show that the proposed RB-Raft algorithm has the ability to resist Byzantine nodes, and its log recognition rate can reach 100%. At the same time, compared with PBFT, the consensus latency of the algorithm in this paper is reduced by 53.3%, and the throughput is increased by 61.8%. The algorithm proposed in this paper is suitable for consensus in untrusted consortium chains. [ABSTRACT FROM AUTHOR]

In order to solve the problems of large number of parameters and high GPU resource cost in binocular vision stereo matching method based on neural network in robot and unmanned driving fields, this paper proposed a multi-scale aggregation stereo matching method. Firstly, this paper proposed a multi-scale feature extraction network to obtain richer features without changing the resolution by using dilated convolution, and introduced the attention mechanism was. Then, features at different resolutions were cross-fused to improve the feature information. Secondly, the acquisition method of cost volume was changed, the cost volume was obtained by aggregation at low scale, and continuously combined the high-scale similar information to update iteratively, and cross fused multiple cost volumes to obtain the final cost volume. Finally, combined with the refinement module of attention mechanism, the outliers and discontinuous regions in the initial disparity map are corrected to obtain the final disparity map. Experimental results show that the algorithm can run under low parameter number and low cost GPU resources, and obtain good matching accuracy. [ABSTRACT FROM AUTHOR]

Motor imagery EEG signals requires a high cost for recording, and there is a large difference for individual’s signals. Cross-subject motor imagery EEG signals recognition task belongs to a typical cross-domain learning problem with a small samples set. To solve this problem, this paper proposed a cross-domain feature learning method for motor imagery EEG signals to improve the recognition performance. The proposed method first selected the optimal measurement to align the covariance of EEG signals, and then extracted common spatial patterns（CSP） from the aligned EEG trials. Second, based on the CSP features, an optimal domain adaptation algorithm was selected to learn the optimal cross-domain features for the target domain. To validate the feasibility and effectiveness of the learned cross-domain features, a classical model was adopted to recognize the learned cross-domain features, and the comparative experiments were conducted on two public datasets. Experimental results show that the learned cross-domain features are obviously better than state-of-the-arts methods in recognition performance. In addition, this paper also compared the parameters setting, performance and efficiency for the proposed method. [ABSTRACT FROM AUTHOR]

In edge computing switching strategy, aiming at the problems of high transmission delay a nd poor environmental adaptability of Markov decision process ( MDP), this paper proposed an edge computing switching strategy integrating fuzzy logic a nd Markov decision process. The method used fuzzy logic algorithm to fuzzify the system parameters, and introduced the fuzzy value into the fitness function to ensure the effective fusion of system parameters. The differential evolution algorithm solved the maximum problem of fitness function, which could select the optimal rules of the environment and improve the adaptability of edge computing to the environment. The algorithm introduced the fitness function into MDP, which improved the comprehensive performance of the system. The scheme took the mobile intelligent device as the initiator of task unloading and the edge server as the object of task unloading. This paper simulated the one -dimensional MDP switching strategy, one-dimensional delay only MDP switching strategy, two-dimensional MDP switching strategy,fuzzy logic MDP switching strategy, minimum distance switching algorithm and minimum delay switching algorithm. The simulation results show that the average task execution time of the edge computing switching strategy of fuzzy logic MDP is 608 . 8 s, which a re 27. 2%,8. 6 %,37 . 1 %,41 % and 22. 3% lower than one -dimensional MDP switching strategy, one-dimensional delay only MDP switching strategy, two-dimensional MDP switching strategy, minimum distance switching algorithm and minimum delay switching algorithm respectively. This scheme not only improves the environmental adaptability of the edge computing switching strategy based on MDP, but also greatly reduces the transmission delay of edge computing. [ABSTRACT FROM AUTHOR]

Targeting to the deficiency of sparrow search algorithm solved the multi-objective problems, and the problems can easily enter the partial optimization and inferior convergence in the counting process, this paper came up with a kind of improved multi-objective sparrow search algorithm. First of all, this paper used a new-type non-dominated sorting to find the Pareto front. Next, it integrated the polynomial mutation and cosine algorithm into species evolution strategy to strengthen its searching ability. It used the species selection method of competition mechanism to decrease the differentiation caused by partial optimized particle and overall optimized particle in the searching process. Finally, this paper compared the improved algorithm and various kinds of multi-objective algorithm in standard test function. The simulation results show that, the convergence and searching ability of the improved algorithm are all superior than other algorithm. Therefore, this algorithm can effectively address the multiple target optimization problem. [ABSTRACT FROM AUTHOR]

For the vehicle routing problem with time window, stochastic travel and service time, and customer classification, this paper constructed a chance constrained programming model, which considered two class customers (ordinary customers and priority customers) and adopted the chance constraint to guarantee the punctual service possibility of priority customers. Meanwhile, this paper designed a variable neighborhood iterated local search algorithm to solve the model, and proposed a minimum waiting time heuristic rule to generate initial solution. Based on Solomon benchmarks, it conducted series of simulation experiments. The results show that the initial solution generation method is valid, the given algorithm can get approximate optimal solutions for both of the deterministic and stochastic problems in a short time, and the customer ratio is positively relative to the number of vehicle usage . The results have a certain reference significance to solve the problem of low customer service level caused by the limited resources and uncertain factors. [ABSTRACT FROM AUTHOR]

Message propagation algorithms from statistical physics are widely used in various fields of artificial intelligence, especially in solving combinatorial optimization problems. According to the related literatures of message propagation algorithm, this paper summarized the history of message propagation algorithm and its related application. According to the development of message propagation algorithm, it introduced the concepts of information propagation algorithm for solving the satisfiability problem, which mainly involved the warning propagation algorithm, the belief propagation algorithm and the survey propagation algorithm. This paper described the convergence and effectiveness research of the three kind of algorithms, summarized the application of each algorithm in related fields, and summarized the research path and application direction of message propagation algorithm. [ABSTRACT FROM AUTHOR]

In recent years, single device sound scene classification has achieved good results, however, progress in multi-device sound scene classification has been slow. To solve the problem of large differences in the number of multi-device samples, this paper proposed a pairwise feature fusion algorithm. By calculating the differences in the spectrogram for each pair of paired samples and averaging these differences after accumulation, obtaining the average spectral characteristics of each device for the conversion of device samples. The algorithm effectively improved the generalization ability of the model while increasing the number of device samples. Meanwhile, in order to obtain global information, it proposed a lightweight attention module, which could make the model focus on the training of the whole sound sequence information on the basis of reduced computation by performing self-attention operations on the input features after compression in the frequency domain. The experimental results show that the proposed algorithm has better advantages in terms of model size and classification accuracy compared with other methods. [ABSTRACT FROM AUTHOR]

As the growth of social network scale, so do challenges to the existing anomaly detection algorithms. Therefore, this paper proposed an anomaly detection method based on graph modularity clustering（GMC＿AD）, which could be applied to solve the problem of low detection efficiency caused by network size and complexity. Based on analyzing the network topology structure, the GMC＿AD method improved the efficiency of events detection by weighting mechanism on abnormal nodes and modularity clustering algorithm. The GMC＿AD processes could be described as follow: a)Since designing a quantization strategy for node evolution in the network, GMC＿AD get the set of abnormal nodes by recognizing nodes with abnormal evolutionary behaviors. b)The method used a modularity clustering algorithm to reduce the network size. c)During the calculation of network fluctuation value, it introduced a weighting mechanism for taking the influence of abnormal nodes into consideration, after that, the GMC＿AD method detected the abnormality by the changes of network fluctuation value. On real social network datasets VAST, EU＿E-mail and ENRON, the GMC＿AD method accurately detected the abnormal periods. The event detection sensibility of GMC＿AD method was increased by 50%～82% meanwhile the run-time efficiency increased by 30%～70%. The GMC＿AD method enhances not only the accuracy and sensitivity but also the efficiency of anomaly detections. [ABSTRACT FROM AUTHOR]

Most of the existing steganalysis methods have weak generalization ability and cannot effectively detect unknown steganalysis algorithms, which makes the accuracy of their classification greatly reduced in the practical application process. To solve this problem, this paper proposed an image steganalysis method based on group convolution and snapshot ensembling（Snapshot Ensembling steganalysis network, SENet）. Firstly, residual convolution block and group convolution block are used to extract the features of the image. Secondly, the model with the best performance is obtained as the snapshot model in each training period. Finally, the selected snapshot models are integrated to classify the images. This method uses the techniques of group convolution and snapshot ensembling to avoid the high training cost of traditional integration methods and the limited generalization ability of a single classifier. Experimental results show that this method can improve the accuracy of steganographic analysis model, and can effectively classify when the training set and test set are mismatched, and has high model generalization ability. [ABSTRACT FROM AUTHOR]

High-quality learning low-dimensional representation of nodes in the graph is a current research hotspot. The existing shallow model methods cannot capture the nonlinear relationship of the graph structure, and the graph convolution model in the graph neural network technology will cause an over-smoothing problem. At the same time, how to determine the role of different hop number relationships in graph representation learning is also a problem that needs to be solved in the research. To solve the above problems, this paper proposes a deep learning model based on T（T＞1） feedforward neural networks. The framework uses deep learning models to extract the nonlinear relationship of the graph structure, and T sub-models effectively capture the local and global（higher-order） relationship information of the graph, and they give different roles in the final vector representation to take advantage of different hop relations. Experimental results on vertex classification and link prediction tasks show that the framework is competitive with existing methods, the benchmark algorithm can be improved by about 20%. [ABSTRACT FROM AUTHOR]

Most sequential recommendation systems（SRSs） assume that the next item to be predicted is related to the user’s previous input. However, in real scenarios, users may click items inconsistent with their own interests and preferences（unreliable instances） by mistake during browsing. This paper proposes a sequential recommendation algorithm based on Gaussian distribution modeling to solve this problem. Firstly, the algorithm extracts the input sequence patterns by reducing the uncertainty of input items through an uncertainty aware graph ensemble network（UAN） with multiple heads of self-attention; Secondly, the extracted input sequence pattern is modeled as a Gaussian distribution, and obtain the dynamic user’s preferences and the uncertainty of preferences in the sequence information; Then, extend the traditional recommended objective function to a sampling loss function and an uncertainty regularizer, and give each training instance appropriate uncertainty; Finally, remove the unreliable examples with high loss and low uncertainty to enhance the accuracy of sequence recommendation. The algorithm is tested on three open datasets, Book Cross, Movielens-1 M and Steam. The results show that the algorithm has achieved an improvement of about 5.3% compared with the baseline with good effects, and has obtained better sequential recommendation results. The proposed algorithm can improve the recommendation accuracy by effectively reducing the uncertainty of input sequence information. [ABSTRACT FROM AUTHOR]

Interactive image segmentation uses a priori information to guide the acquisition of the people ' s interesting parts of an image, but existing algorithms cannot balance efficiency and accuracy. To solve this problem, this paper proposed a fast interactive segmentation algorithm based on superpixel and random walk (random walk on superpixel, SPRW) . Firstly, it presegmented the image into superpixel regions with local similarity and used pixel colour averages to represent the superpixel regions. Secondly, it built the F-B graph structure based on manually labelled a priori information and extended the range of the random walk, obtained the hard segmentation results using the random walk solution. Finally, this paper proposed the improved matting algorithm (fast robust matting, FRB ) for the unsmooth boundaries of segmentation results and performed the secondary process to obtain soft segmentation results. Experiments on the BSDSOO and MSRC datasets confirm that the proposed hard method is superior to other algorithms in terms of time and average intersection ratio, while experiments on the Alpha Matting dataset demonstrate that the proposed soft algorithm improves efficiency and accuracy. In addition, the experiment of replacing the background of a life photograph demonstrates the application value of the proposed algorithm. [ABSTRACT FROM AUTHOR]

To solve the problem that existing botnet detections targeted a single phase of the botnet lifecycle, this paper proposed an online botnet detection method based on ensemble learning. Firstly, this paper fine-grained labeled the traffic of multiple phases of botnet lifecycle to generate a botnet dataset. Secondly, this paper combined multiple feature selection algorithms to generate a significant feature set containing 23 features and a less significant feature set containing 28 features. It integrated multiple deep learning models based on stacking ensemble learning and provided different input feature sets for different primary class ifiers to obtain a botnet online detection model. Finally, this paper deployed the botnet online detection model to detect multiple botnets online at the network entrance. Experiment shows that the proposed botnet online detection method based on ensemble learning in this paper can effectively detect multiple stages of botnet traffic, and the malicious traffic detection rate can reach 96 . 47% . [ABSTRACT FROM AUTHOR]

This paper proposed a natural language sentence matching method that combined high-level and low-level semantic information to solve the problems about current natural language sentence matching method fail to integrate common semantic information and it is difficult to capture deep-semantic information. First of all, the method used pre-trained word vector GloVe and character-level word vector to obtained the word embedding representation of sentence P and sentence Q. Secondly, this paper encode red P and Q with bidirectional LSTM, then it contained low-level semantic information through preliminary fusion of P and Q. Thirdly, this paper calculated bidirectional attention between P and Q, then spliced them together to get semantic representation, afterwards it calculated its self-attention to obtained high-level semantic information. Finally, this paper used a heuristic fusion function to fuse the low-level semantic information with the high-level semantic information to obtain the final semantic representation, and it used a convolutional neural network to prediction answers. This paper evaluated the proposed model on two tasks, such as recognition textual entailment, paraphrase recognition. This paper conducted experiments on the SNLI dataset and the Quora dataset. The results show that the accuracy of the proposed algorithm on the SNLI test set is 87. 1%, and the accuracy of the Quora test set is 86. 8%, which verifies the effectiveness of the algorithm in the task of natural language sentence matching. [ABSTRACT FROM AUTHOR]

Network representation learning can effectively solve the problem of data sparsity in recommendation. In this paper, LINE and DeepWalk in network representation learning were improved, and a hybrid recommendation algorithm was proposed to be applied to movie recommendation scene. The new algorithm generates three low dimensional feature vectors by learning user preference feature, user aversion feature and similar user feature. Three low dimensional feature vectors are linearly combined to form a user representation vector, and cosine similarity is used as the similarity index to recommend the movies associated with similar users to target users. Experimental results show that, compared with the suboptimal algorithm, the accuracy and Fl index of the proposed algorithm are improved by 12% and 7% respectively on MovieLens dataset, and 16% and 18% respectively on MovieTweetings dataset. The recommendation algorithm based on multi-dimensional feature representation learning proposed in this paper has significant advantages in movie recommendation scenes. [ABSTRACT FROM AUTHOR]

Aiming at the problems of parallel support vector machine (SVM) algorithm in big data environment such as redundant data sensitivity, difficulty in parameter selection, and low parallelization efficiency, this paper proposed a parallel SVM algorithm using Relief and bacterial foraging optimization (BFO) algorithm based on MapReduce (RBFO-PSVM). Firstly, the algorithm designed a feature weight calculation strategy (MI-Relief), which used mutual information to improve the weight calculation function of Relief algorithm to eliminate redundant features in the data set and effectively reduce redundant data to support parallelism. Secondly, this paper proposed a hybrid BFO algorithm based on MapReduce (MR-HBFO), which selected the optimal parameters of SVM in parallel, and solved the problem of difficult selection of SVM parameters. Finally, it proposed the kernel clustering strategy (KCS) to reduce the size of the data set involved in parallel training, and proposed a cross-fusion cascaded parallel SVM (CFCPSVM) to improve the cascade SVM (CSVM) feedback mechanism. It trained SVM by combining with the Map Reduce programming framework, and this improved the parallelization efficiency of parallel SVM. Experiments show that the RBFO-PSVM algorithm has a better classification effect on large data sets and is more suitable for large data environments. [ABSTRACT FROM AUTHOR]

In order to improve the safety and smoothness of flight, and solve the problems of excessive turning angle and too many search path nodes in the traditional A• algorithm, this paper proposed a synchronous bi-directional A• search algorithm based on sector field expansion. Firstly, this paper extended the hazardous area boundaries based on the raster map method. Secondly, it designed the A• algorithm based on synchronous bi-directional search, which dynamically defined the target nodes for forward and reverse search. Aiming at the problem of limiting search angles, this paper proposed a sector expansion strategy in 5 x 5 domain, and designed an evaluation function with double weight parameters to reduce the generation of redundant points. Finally, in order to verify the effectiveness of the improved algorithm, this paper selected square and irregular shape hazard areas for simulation. The results show that the improved algorithm searches for a smoother path. Compared with the results of the traditional bi-directional A• algorithm, the length of its search path reduce by 1. 65% and 13. 16%, and the number of its search path nodes reduce by 42. 6% and 46. 81 %, respectively, under different shapes of hazard areas, which has a strong search efficiency. [ABSTRACT FROM AUTHOR]

In order to solve the problem of collision free path planning and system efficiency improvement of multi AGV in dynamic unstable environment, this paper proposed a collision free path planning method based on time window for AGV. Firstly, this paper established a multi-AGV collision avoidance model, combined with the time window model, and divided the collisionfree path planning of multiple AGV into two stages: pre-planning and real-time planning. In the pre-planning stage, this paper calculated the conflict-free time window of multiple AGV and maximized the circulation of AGV in the system. In the real-time planning stage, this paper used the method of changing the AGV' s occupancy priority on the collision avoidance model and local re-planning to dynamically avoid collisions. Finally, a simulation experiment with a smart warehouse as an application case proves that the algorithm can effectively avoid the collision of multiple AGV, increase the circulation of AGV, and has good robustness and flexibility in a dynamic environment. [ABSTRACT FROM AUTHOR]

Single-view 3D reconstruction based on deep learning is a research hot spot at present. In order to discover more high-frequency details, SDF-SRN algorithm introduces positional encoding, but neural network is easy to overfit without accurate supervision, and reconstructs uneven surface. To solve the problem, this paper proposed the network model based on sparse feature. The model enabled the network that preferred to overfitting to predict high-frequency residual by residual learning. The feature extraction network extracted sparse features and the global features. Then one hypernetwork took the sparse features as input and generated prediction shallow head. This shallow head predicted low-frequency part of signed distance function. Another hypernetwork took global features as input and generated another shallow head. This shallow head predicted high-frequency residual. It fused two predictions of shallow heads into final signed distance function. Spectrum analysis shows that the design purpose of network is achieved. Compared with other smooth surface reconstruction schemes, the network can achieve smoother surface reconstruction with enough details. It overcomes the overfitting of SDF-SRN. The qualitative and quantitative comparison with other advanced single-view reconstruction approaches show the superiority of the proposed approach. [ABSTRACT FROM AUTHOR]

In recent years, malware is constantly developing and changing, resulting in a low accuracy of a single detection model. Using ensemble learning to combine multiple models can improve the detection effect, but the accuracy and diversity of base learners in the ensemble model is difficult to balance. Therefore, this paper proposed an ensemble model generation method based on genetic programming to generate the optimal ensemble model for malware detection. Genetic programming could integrate feature processing and ensemble model construction into a single GP tree, which solved the problem that traditional malware ensemble detection models was difficult to balance individual accuracy and diversity. The method used the detection rate of malware in the ensemble model as the basis of population evolution to ensure the accuracy of the base learner. When generated an ensemble model, it could automatically select the feature processing method, classification algorithm and optimize the hyperparameters of the base classifier, and ensured the diversity of the base learner through the perturbation of input attributes and algorithm parameters. According to the idea of survival of the fittest, it evolved to generate the optimal ensemble model with high accuracy and diversity. The experimental results on the EMBER dataset show that the detection accuracy of the best ensemble model reaches 98. 88%. Further analyses show that the model has high diversity and interpretability. [ABSTRACT FROM AUTHOR]

To solve the problems of low convergence accuracy and easy to fall into local optimum in the optimization process of the original artificial bee colony algorithm, this paper proposed an enhanced artificial bee colony algorithm with strong and weak ties theory (SWT-ABC). Firstly, it defined the strong and weak ties model in a matrix multi-groups structure and stipulated that the strong ties individuals randomly guided the search process from three directions to improve the convergence speed and accuracy. In order to enhance the ability to jump out of the local optimum, SWT-ABC adopted the interaction of weak ties individuals among subgroups to enhance the population diversity. Finally the algorithm added the reverse learning mechanism of scout bees and determined appropriate upper limit of honey source to improve the utilization efficiency of the evaluation times. Through the numerical experiment of benchmark functions and comparisons with 12 improved intelligent optimization algorithms, it shows that the SWT-ABC has high convergence accuracy and good robustness, and also gets good performance in solving high-dimensional optimization problems. [ABSTRACT FROM AUTHOR]

To address the problems of Byzantine fault-tolerant algorithm with high communication overhead, simple node selection, and lack of penalty mechanism for malicious nodes, this paper proposed an improved Byzantine fault-tolerant consensus algorithm based on the recommendation trust model. It introduced the recommended trust model under P2P networks to calculate the global trust value of each node based on the behavior of nodes in the consensus phase, and used the node selection mechanism is to solve the problem of simple node selection. The nodes with high global trust value entered the consensus group, and malicious nodes were kicked out of the consensus group and no longer participate in the consensus, solving the problem of lack of punishment mechanism for malicious nodes. Experiments show that the R-PBFT algorithm has lower network overhead and higher fault tolerance than the PBFT algorithm. [ABSTRACT FROM AUTHOR]

The deep decision tree migration learning boosting method ( DtrBoost) can effectively realize the migration learning from a single source domain to a target domain under supervision, but can not realize the unsupervised migration scenario under multiple source domains. To solve this problem, this paper proposed an unsupervised transfer learning boosting method for optimizing the weight under multi-source domain distribution. The main idea was to calculate the corresponding KL value according to the distribution of different source domains and target domains, selected an appropriate number of samples from different source domains to train the classifier and pseudo label the samples from the target domain. Finally, the algorithm assigned different learning weights according to the KL distance of each different source domain, and the labeled source domain samples integrated to the pseudo labeled target domain to obtain the final result. Comparative experiments show that the proposed algorithm achieves better classification accuracy and adaptive effect on different data sets. The average classification error rate decreases by 2. 4% and more than 6% on the best marketing data set. [ABSTRACT FROM AUTHOR]