Search

Traditional federated learning faces challenges such as high communication costs, structural heterogeneity,and insufficient privacy protection. To address these issues, this paper proposes a federated learning evolutionary algorithm that applies sparse evolutionary training algorithm to reduce communication costs and integrates local differential privacy protection for participants’ privacy. Additionally, it utilizes the NSGA-Ⅲ algorithm to optimize the network structure and sparsity of the global federated learning model, adjusting the relationship between data availability and privacy protection. This achieves a balance between the effectiveness, communication costs, and privacy of the global federated learning model. Experimental results under unstable communication environments demonstrate that, on the MNIST and CIFAR-10 datasets, compared to the solution with the lowest error rate using the FNSGA-Ⅲ algorithm, the proposed algorithm improves communication efficiency by 57. 19% and 52. 17%, respectively. The participants also achieved（3. 46, 10-4） and（6. 52, 10-4）-local differential privacy. This algorithm can effectively reduce communication costs and protect participant privacy without significantly compromising the accuracy of the global model. [ABSTRACT FROM AUTHOR]

In view of the problems of data sparsity and the diversity in academic paper recom-mender systems,based on CONVNCF, this paper proposed an algorithm of convolution with word and heterogeneous information network for academic paper recommendation ( WN -APR) . Firstly, WN -APR algorithm learned user and paper' s diverse features from different semantics to alleviate the sparsity problem. Then it designed an outer product fusing way to seamlessly combine user features with paper features. Replacing of 2D CNN, this algorithm applied 3 D convolution to mine the influence of different features on the performance. Finally, it modified the BPR loss function to enhance diversity in recommendations. Experimental results on CiteULike-a and CiteULike-t datasets show that WN-APR improves the performance of accuracy and diversity over the baseline models. [ABSTRACT FROM AUTHOR]

There are many problems in the process of collecting paper data using single data source, such as insufficient data comprehensiveness and limited data collection speed due to website access frequency limitation. Aiming at these problems, this paper proposed a paper data crawling technology for multi-data sources. Firstly, it used the four Chinese document service websites-How Net, Wanfang Data, Weipu, and Chaoxing as data sources, completed the task of crawling and parsing list page data for the search keywords. Then it used the task scheduling strategy to remove repeated data and balance the tasks. Finally, it used multi-threads for each data source to crawl, parse and store the detail information of the papers, and built a website for search and display. Experiments show that under the same crawling and parsing speed, this technology can complete the paper information collection task more comprehensively and efficiently, which proves the effectiveness of this technology. [ABSTRACT FROM AUTHOR]

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 a collaborative topic regression model based on the preference for frequent topic sets to address the item-cold-start problem in academic paper recommendation. The algorithm took into account the user' s preference for research hotspots when selected academic papers, and used frequent topic sets to represent research hotspots. So, it expressed user' s preference for research hotspots as the user's preference for frequent topic sets. Firstly, it obtained the papers-topic probability distribution matrix through LDA algorithm and filter out the topics with higher probability. Then, the algorithm mined the frequently-occurring topic sets and gets the relationships between papers and frequent topic sets. Finally, it used the user's preference for frequent topic sets for the prediction of unknown scores. Experiments on CiteULike datasets show that the algorithm improves the recall, accuracy and RMSE over the matrix factorization model and the collaborative topic regression model. [ABSTRACT FROM AUTHOR]

Considering the unique users of scholars' social networks, this paper proposed a collaborative association model of paper and project data based on scholars' social networks. Firstly, the proposed model used the two-step feature selection method to preprocess the data, removed the irrelevant and redundant features and obtained the effective features that affected the association between the paper and the project. Then it would adopt TVSM to calculate the text similarity between the paper and the project, then formed recommendation sets for different papers/projects. Through the social network (SCHOLAT) data for researchers, the model was implemented and applied to SCHOLAT. The online application situation and user feedback show that the model has good accuracy and practicability. Furthermore, it can more fully explore the potential relationship between the paper and the project, provide users with better academic research management services, and propose a novel research method for analyzing the academic big data. [ABSTRACT FROM AUTHOR]

In view of the problem of fuzziness of rating and tag in traditional collaborative filtering algorithms,this paper used trapezoidal fuzzy number to describe the mapping relationship between rating and satisfaction.The algorithm considered the impact of sparseness of the rating,constructed a new trapezoidal fuzzy rating model to determine the similarity based on fuzzy rating,analyzed the degree of membership between the tag and the item,and constructed a fuzzy item-tag matrix to measure the similarity based on the degree of tag membership.Finally,it used the improved scoring prediction strategy to estimate the score.The experimental results on the MovieLens dataset show that the proposed algorithm improves the prediction accuracy while suppressing the cold start of the project,alleviating the problems of fuzziness and sparseness,which indicates the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Aiming at the low efficiency caused by massive academic paper data, this paper proposed a recommendation algorithm method based on the hierarchical mixed model named WSVD ++. According to the structural features of academic papers, the model constructed a weighted bipartite graph model. Firstly, this method extracted the features of each paper and constructed the composite relation graph according to the ratio of different features. Secondly, it used an improved PPR algorithm on the graph to calculate the importance weight of each paper, and then weighed the relation between the user and the paper. Finally, it recommend on the weighted bipartite graph by using SVD + + graph algorithm. The result shows that the proposed algorithm improves the recommended accuracy. The whole process implemented in distributed graph calculation system, that means the method has good expansibility and is suitable for big data processing. [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]

The current 3D-ConvNet-based action recognition algorithms generally use GAP to compress feature information. However, it leads to issues of information loss, redundancy, and network overfitting. To address these issues and enhance the retention of high-level semantic information extracted by the convolutional layer, this paper proposed an action recognition al- gorithm based on GFDP. Firstly, DCT shows that GAP is a special case of feature decomposition in the frequency domain. Therefore, the algorithm introduced more frequency components to increase the specificity between feature channels and reduce the information redundancy after information compression. Secondly, to better suppress the overfitting problem, the algorithm introduced the batch normalization strategy to the convolutional layer and extended it to the fully connected layer of the action recognition model with ERB-Res3D as the skeleton to optimize the data distribution. Finally, this paper verified the proposed method on the UCF101 dataset. The results reveals that the model's computational load is 3.5 GFlops, with 7.4 million para- meters. The final recognition accuracy improved by 3.9% based on the ERB-Res3D model and 17.4% based on the original Res3D model. This improvement effectively achieves more accurate behavior recognition results. [ABSTRACT FROM AUTHOR]

Existing NL2SQL approaches do not fully utilize the information of data table columns, which plays an important role in the semantic understanding of the problem and the logical generation of SQL statements. This paper proposed an NL2SQL method that fused data table columns (FC-SQL) to improve the overall accuracy of SQL generation. Firstly, this method utilized BERT to merge the problem and database table columns for encoded representations. Secondly, it used multitask learning approach to construct a multi-task network by combining parallel and cascade to predict different sub-tasks. Finally, for the conditional value extraction sub-task, this method computed the similarity between the words in the problem and the table columns by fusing the information of the columns, and it used the similarity value as a weight to compute each word as the conditional value probability of each word as a conditional value, thus improving the accuracy of conditional value prediction. The logical form accuracy and SQL execution accuracy on the TableQA dataset reach 88.23% and 91.65%, respectively. This paper designed ablation experiments to verify the effect of table columns information on the model. The experimental results show that the incorporation of table columns improves the effectiveness of the conditional value extraction sub-task, which in turn improves the overall accuracy of the NL2SQL task and provides better SQL generation compared to the comparison model. [ABSTRACT FROM AUTHOR]

The current methods for short-time precipitation nowcasting are based on radar echo extrapolation model, without fully considering the close influence of other meteorological factors on the evolution of precipitation generation and cancellation, thus limiting the accuracy of the forecasts. To address the above issues, this paper produced a short-time precipitation nowcasting dataset, and proposed the MFPNM(multiple factors precipitation nowcasting model). Based on data from the Fengyun4B satellite, the dataset toke quantitative precipitation estimation as the forecast object and contained four background meteorological factors. Taking the TransUNet as the backbone of the model, this model proposed the parallel dual encoder to extract the high-dimensional spatio-temporal features of the forecast object and the background meteorological data, respectively. Besides, it constructed the content coding module to encode the spatial features of the background data as the learnable positional embedding of the high-dimensional feature vectors of the forecast object. It used a Transformer module to construct the global relationship between the high-dimensional features of the sequence data for better sequence prediction. The metrics used in this paper included critical success index, false alarm rate, root-mean-square error, and structural similarity, etc. The MPFNM was evaluated on two datasets (the proposed dataset and an open-source dataset) and outperformed the baseline models, and it was analyzed for explainability through the SHAP technique. The experimental results and explainability analysis show that the model has better forecasting accuracy and reliability. [ABSTRACT FROM AUTHOR]

The existing branch prediction models can't predict the behaviors of various instructions in the processor accurately, which leads to the limitation of processing efficiency. Therefore, this paper proposed two hybrid prediction solutions, which aimed to combine multiple branch prediction models to improve prediction accuracy and processor execution efficiency. This paper transferred the prediction results of TAGE branch prediction model and BATAGE branch prediction model to the Hybrid model. In the prediction phase, the Hybrid model selected the prediction results of the best-performing branch prediction model based on the historical performance of TAGE and BATAGE. In the update phase, the Hybrid model updated the saturation counter of the entries that needed to be updated according to the designed hybrid prediction strategy. Experiments on 440 test programs provided by the CBP software simulation platform show that the prediction error rates of both hybrid prediction solutions are lower than those of many of the latest mainstream branch prediction models. This study provides an effective solution to the problem of predicting all command mode behaviors. In the branch instruction prediction of real CPU, this research provides some practical value. [ABSTRACT FROM AUTHOR]

Aiming at the distributed unrelated parallel machine scheduling problem with machining constraints, this paper proposed an extended empire competition algorithm to solve the problem, which took the total transportation cost, evaluation function for simultaneous shutdown of parallel machines among factories and equilibrium evaluation function for average switching frequency of workpiece types among factories as optimization objectives. Based on the original empire competition algorithm, this algorithm added an initial factory-workpiece sequence group suitable for factory assignment. According to the shortcoming of the traditional empire competition algorithm that it was easy to fall into the local optimum, this paper divided the inferior sequence assimilation into the external assimilation mechanism and the internal assimilation mechanism. The extended empire competition algorithm combined local and global search methods to realize the intelligent search behavior. This paper used the partial matching crossover and single point mutation to update the factory-workpiece sequence group and ensure the diversity of the factory-workpiece sequence. Finally, this paper designed three different scales of 12 examples and verified the effectiveness of the proposed algorithm through simulation experiments. At the same time, this paper verified the advantages of the proposed algorithm in solving the distributed multi-objective unrelated parallel machine scheduling problem by comparing the research results in related fields. [ABSTRACT FROM AUTHOR]

Aiming at the floorplanning problem with only hard modules, it is usually constructed as a combinatorial optimization model, but the time cost of solving process is high. In order to improve the solving efficiency, this paper proposed a floorplanning algorithm based on non-smooth analytic mathematical programming. Based on the coordinate representation of modules, this paper established a non-smooth mathematical programming model, which was a generalization of the optimization models corresponding to various optimization stages of different cases that were solved by the conjugate sub-gradient algorithm (CSA). Aiming at the fixed-outline floorplanning problem, this paper achieved wirelength optimization under fixed-outline constraint through the general framework consisted of three stages: global floorplanning, legalization and local optimization. To address the case without fixed-outline constraint, this paper proposed a conjugate sub-gradient algorithm with golden section strategy (CSA_GSS). The algorithm adopted golden section strategy to reduce the area of the fixed-outline and to achieve the dual effect of both area and wirelength optimization. Compared with floorplanning algorithm based on B-tree on GSRC test circuit, the proposed algorithm had advantages in terms of wirelength and time for large-scale circuits. Experimental results show that the algorithm can obtain better wirelength with lower time complexity. [ABSTRACT FROM AUTHOR]

The survival behavior of snakes in nature generates the snake optimization (SO), a meta-heuristic optimization algorithm. But the original snake optimization algorithm suffers from slow convergence and easy to fall into the local optimum, so this paper proposed an improved snake optimization algorithm that combined elite initialization and K-nearest neighbors improved snake optimizer (EKISO). Firstly, in order to improve the quality of the initial population, it proposed elite initialization during the population initialization stage, which generated high-quality initial population individuals based on elite individuals of the population. Secondly, optimizing the spiral foraging strategy with a shock factor expanded the search range during the global exploration stage and enhanced the algorithm's local escape capability. Finally, in the local exploitation stage, it proposed a position update method based on the K-nearest neighbor concept, enhancing the information interaction capability among population individuals, thus accelerating convergence speed and improving convergence accuracy. This method compared with seven other optimization algorithms using 14 classical test functions and 4 CEC2017 test functions, which proves that EKISO converges faster, has higher accuracy and is not easy to fall into local optimum. To further validate the practicality and feasibility of EKISO, this paper applied it to pressure vessel design problems. Experimental comparative analysis reveals that EKISO possesses certain advantages in dealing with real optimization problems. [ABSTRACT FROM AUTHOR]

In view of the fact that there is still a certain gap in representing and reasoning with the orientation relation between real objects due that the existing models for representing and reasoning with direction relations use a point and minimum bounding rectangle to approximate the spatial object itself, this paper proposed a new model for dealing with the disconnected spatial objects based on Voronoi diagram. With the help of the theory of Gestalt psychological, the proposed model constructed visual regions between spatial objects by extracting the feature points and feature chains of the disconnected spatial objects to generate the direction Voronoi diagram, which realized the representation of directions between the space objects which were disconnected and contain holes. This model fully considered the influence of the shape, size and other factors of the spatial objects, and had higher expression accuracy and wider applicability. In order to improve the precision of reasoning with cardinal directional relations between complex spatial objects, this paper proposed an algorithm for composing cardinal direction relations defined by the proposed model, which reduced the uncertainty of the results of composition by means of Tile-union operation and Pr operation. The results of analysis and verification show that the proposed model and algorithm reduce the uncertainty of the reasoning results, improve the accuracy of representation and reasoning, and then enhance the ability to analyze and process the direction relations with complex spatial objects. [ABSTRACT FROM AUTHOR]

The key to multimodal sarcasm detection is effective to align and fuse the features of different modes. However, the existing multimodal data fusion methods ignore the relationship between multimodal intercomponent structures. Also, the importance of common features associated with sarcastic emotions in multimodal data is overlooked in the process of recognizing sarcasm. To address the above problems, this paper proposed a model based on cross-modal hierarchical interaction networks and contrastive learning (CMHICL). Firstly, the cross-modal hierarchical interaction network employed a minimal unit alignment module based on the cross-attention mechanism and a compositional structure fusion module based on the graph attention network to identify inconsistencies between text and images at different levels, and determined the samples with low consistency as sarcasm samples. Secondly, two contrastive learning tasks, based on data enhancement and category enhancement, helped to learn common features related to sarcasm and reduce false correlations within the modality. The experimental results show that the CMHICL model has increased the Acc by 0.81% and the F₁ value by 1.6% compared to the baseline models, which verifies the key role of the hierarchical interactive network and contrastive learning method proposed in this paper in multimodal sarcasm detection. [ABSTRACT FROM AUTHOR]

Knowledge reasoning is a fundamental task in knowledge graph completion. It is a popular topic in the academie community. Integrating reinforcement learning and knowledge reasoning is a viable solution to improve the inference effectiveness and interpretability of models. Taking the problem of knowledge graph research as a path or sequence decision problem in the knowledge reasoning method based on reinforcement learning can make better use of semantic information, such as entities and relationships, to improve reasoning effect and interpretability. This paper provided a descriptive overview of the basic concepts of knowledge graph and knowledge reasoning, and described the research progress in recent years. The paper analysed and compared the related research on knowledge reasoning based on reinforcement learning from two perspectives; single-layer and double-layer reinforcement learning knowledge reasoning. Furthermore, the paper explored the application of knowledge reasoning in various domains, including knowledge question answering, intelligent recommendation systems, healthcare, and transportation. Lastly, the paper discussed future research tendencies for reinforcement learning-based knowledge reasoning and offered insights into potential avenues for exploration and development. [ABSTRACT FROM AUTHOR]

The federated learning framework that uses autoencoder model to detect malicious model updates is an excellent defense framework for poisoning attacks. However, the existing autoencoder-based models face several challenges such as training difficulties and limited anomaly detection capability. In view of the above problems, this paper proposed an algorithm for detecting malicious model updates of federated learning based on β-VAE: the central server stabilized the training dataset by suppressing random attributes of the training samples, and used the dataset to train β-VAE anomaly detection model in real time. The model computed the anomaly score of task model updates uploaded by the clients, and then to detect and exclude malicious model updates based on the dynamic threshold of anomaly score. This paper evaluated the performance of the proposed algorithm on three federated learning tasks. Specifically, these tasks include classification on the MNIST dataset using the logistic regression (LR) model, classification on the FEMNIST dataset using the convolutional neural network (CNN) model, and character prediction on the Shakespeare dataset using the recurrent neural network (RNN) model. The experimental results show that, the task model under this algorithm achieves higher accuracy compared to other defense algorithms. This indicates that in Non-IID scenarios, the algorithm exhibits strong robustness against poisoning attacks in federated learning. [ABSTRACT FROM AUTHOR]

Aiming at the problem that the positioning accuracy of traditional SLAM algorithm decreases due to the influence of dynamic feature points in dynamic environment, this paper proposed SF-VINS which fused semantic information. Firstly, it incorporated the PP-LiteSeg semantic segmentation network into the front-end of system based on the VINS-Mono algorithm and removed the dynamic feature points according to the results of semantic segmentation. Then, it constructed the semantic probability error constraint with the pixel semantic probabilities and employed adaptive weights for feature points in the backend. Based on these modifications, this paper proposed a new BA cost function and an optimization strategy for camera extrinsic parameters to improve the accuracy of state estimation. Finally, this paper conducted experiments on the VIODE and NTU VIRAL datasets to validate its effectiveness. The experimental results show that the proposed algorithm has advantages in terms of positioning accuracy and robustness in both dynamic and static scenes compared with state-of-the-art visual-inertial SLAM algorithms. [ABSTRACT FROM AUTHOR]

Existing cost allocation methods for adaptive video steganography mainly focus on specific transform coefficients, resulting in lower capacity. Moreover, distortion drift is a significant challenge for steganography in HEVC videos. Therefore, this paper proposed a cost allocation method that combined the intra-frame and inter-frame processes of HEVC video coding to achieve high-capacity, low-distortion transmission in high-performance adaptive video steganography. Firstly, the method investigated the discrete sine transform features in HEVC video coding, analyzing the error propagation patterns of these coefficients under disturbance. During the embedding processed, it conducted a detailed analysis on the intra-block distortion, interblock distortion, and inter-frame distortion caused by modifying transform coefficients. The algorithm also took into account the differentiation in inter-block distortion resulting from steganography in different blocks, leading to the classification of blocks. This paper maximized the utilization of all non-zero transform coefficients, allocating distinct distortion costs for various carrier coefficients. The covert information was then embedded into frames that minimally impacted video quality. Experimental results indicate that, compared to existing HEVC video coefficient domain steganography methods, the proposed algorithm demonstrates advantages in terms of video bitrate, video quality, and embedding capacity. [ABSTRACT FROM AUTHOR]

To solve the problems of low recognition accuracy and poor generalization performance of existing face anti-spoofing models in different application scenarios, this paper adopted the idea of disentangled representation learning and proposed a face anti-spoofing method based on disentangled representation learning. This method adopted U-Net architecture and ResNet18 as the encoder-decoder. In the first stage of training, it inputted real samples so that the encoder only learned information related to real samples. In the second stage, this paper built an adversarial learning network, inputted samples without labels, feature fusion of the pre-trained encoder output and the new encoder output, reconstructed the image by the decoder, and performed adversarial training with the original image in the discriminator to achieve feature decoupling. Compared with some classic face anti-spoofing methods, the model paper achieved better detection performance. The lowest detection error rate in several experiments on the OULU-NPU data set is only 0.8%, which is better than classic detection methods such as STDN. The face anti-spoofing method used staged training to enable the model to obtain a more discriminative feature representation than the end-to-end model in adversarial training. It adopted a multi-classification strategy in the deception feature map output stage to reduce the impact of different image noises on classification results, and experiments on public data sets verified the effectiveness of the algorithm. [ABSTRACT FROM AUTHOR]

LR serves as a generalized linear classifier for binary classification in supervised learning, exhibiting characteristics of simplicity in structure, strong interpretability, and effective fitting when dealing with linear data. However, its classification performance becomes limited when confronted with high-dimensional, uncertain, and linearly inseparable data. To address the inherent limitations of logistic regression, this paper introduced the theory of granular computing and proposed a novel logistic regression model called rotating granular logistic regression (RGLR). This paper introduced the theory of rotating granulation, where different angles of rotation were applied to pairs of features forming a plane coordinate system. This process constructed rotating granules by rotating pairs of features at various angles on the plane coordinate system, and granulated to form rotating granule vectors on multiple plane coordinate systems. This paper further defined the size, measurement, and operational rules of granules, and proposed a loss function for rotating granular logistic regression. The optimized solution of the rotating granular logistic regression was obtained by solving the value of the loss function. Finally, experiments are conducted using multiple UCI datasets, and the results compared across various evaluation metrics, indicate the effectiveness of the rotating granular logistic regression model. [ABSTRACT FROM AUTHOR]

To overcome AMR s limited generalization and low classification accuracy in non-cooperative communication contexts with low signal-to-noise ratio, this paper proposed a model named CGDNN, which integrated CNN, GRU and deep neural networks. To mitigate noise impact on modulation detection, this paper initially denoised I/Q sampled signal using wavelet thresholding. Subsequently, this paper utilized CNN and GRU for extracting spatial and temporal features from signals before proceeding to classification through fully connected layers. Besides enhancing AMR performance, the CGDNN model significantly reduced computational complexity compared to competitors. Experiment results demonstrate an average recognition accuracy of 64.32% on the RML2016. 10b dataset, with an enhancement in signal classification accuracy from 12 dB to 0 dB. Moreover, the model substantially decreased confusion between 16QAM and 64QAM, achieving a peak recognition accuracy of 93.9% at 18 dB. CGDNN model effectively improved AMR detection accuracy under low signal-to-noise ratio conditions and enhanced model training efficiency. [ABSTRACT FROM AUTHOR]

Most of the existing MAPF methods based on communication learning have poor scalability or aggregate too much redundant information, resulting in inefficient communication. To solve these problems, this paper proposed disruptor identifiable communication (DIC), which learned concise communication excluding non-disruptors by judging whether the agent in the center of the field of view would change its decision-making due to the presence of neighbors, and successfully filtered out redundant information. At the same time, this paper further instantiated DIC and developed a new highly scalable distributed MAPF solver: disruptor identifiable communication based on reinforcement and imitation learning algorithm (DICRIA). Firstly, the disruptor discriminator and the policy output layer of DICRIA identified the disruptor. Secondly, the algorithm updated the information of the disruptor and the communication wish sender in two rounds of communication respectively. Finally, DICRIA output the final policy according to the coding results of each module. Experimental results show that DICRIA S performance is better than other similar solvers in almost all environment settings, and the algorithm increases the success rate by 5.2% on average compared to the baseline solver. Especially in dense problem instances with large-size maps, the algorithm even increases the success rate of DICRIA by 44.5% compared to the baseline solver. [ABSTRACT FROM AUTHOR]

To solve the flexible job shop dynamic scheduling problem with fuzzy quality control time, which is more realistic, this paper established two rescheduling models for machine breakdown and urgent order insertion with the objective of minimizing completion time. Two models were for three failure scenarios: emergency order insertion, machine failure during no-load operation, and machine failure during work-piece processing. To solve this model, this paper designed a novel genetic-neighborhood search algorithm that integrated cellular-automata-based neighborhood search and random restart hill-climbing. Aiming at the characteristics of dual sub-decisions involving order sequencing and machine selection, this paper designed a two-layer coding scheme including process code and machine code, and designed corresponding genetic operations such as crossover and mutation for them. After that, this paper applied the genetic operations into the framework of cellular-automata-based neighborhood search to enhance the algorithm's global search capability, and integrated the random restart hill-climbing algorithm based on key operations to improve the algorithms local development capability. Experiments tested 10 flexible job shop scheduling instances and one flexible job shop scheduling instance with fuzzy quality control time to verify the effectiveness of the proposed algorithm and models. In addition, the test results under different failure scenarios show the proposed method outperforms the backward-based scheduling strategy used in practice. [ABSTRACT FROM AUTHOR]

In order to solve the problem that the daily rehabilitation training of cardiovascular patients depends on directing by health care professions in rehabilitation centre, during assessing and correcting the movements system about cardiovascular patients independent rehabilitation training at home, this paper proposed an action assessment method for cardiovascular patients' rehabilitation training based on key points of the human skeleton (ASRT-PHS). Firstly, this paper constructed a dataset for rehabilitation training actions using a camera and data augmentation in accordance with the specified rehabilitation training specification for cardiovascular patients. Secondly, this paper employed a deep learning-based detector and pose estimator to capture human body positions and extract key points of the human skeleton, respectively, and then input the results into a convolutional neural network for action recognition. Thirdly, by calculating joint angle thresholds, joint distance ratio and assessing standard motions, this paper constructed a motion segmentation model based on joint distance ratios and an action assessment model based on action joint angle thresholds. This paper investigated the optimal combination of ASRT-PHS by assessing its performance with various joint angle thresholds and action recognition approaches. The results show that ASRT-PHS achieves an average action recognition, segmentation and assessment accuracy of 92.78%, 77.6% and 87%, respectively. Furthermore, case tests about the true cardiovascular patients show that the average accuracy of the prototype system is 71.3%, which provides a feasible intelligent auxiliary system for patients autonomous rehabilitation training at home. [ABSTRACT FROM AUTHOR]

Maintaining population diversity and improving algorithm search efficiency are two major problems that need to be solved urgently in the multimodal multi-objective optimization. To address the above problems, this paper proposed a multimodal multi-objective brain storm optimization algorithm based on zoning search and reinforcement learning (MMBSO-ZSRL). In the MMBSO-ZSRL, the decision space was first decomposed into multiple subspaces to reduce the search difficulty and maintain the population diversity. Subsequently, the proposed algorithm used SARSA algorithm to balance the global exploration and local exploitation capabilities of the brain storm optimization algorithm. Additionally, the MMBSO-ZSRL utilized the special crowding distance to select individuals for guiding the population evolution. To verify the performance of the proposed algorithm, this paper selected six advanced multimodal multi-objective optimization algorithms and the IEEE CEC2019 multimodal multi-objective problem benchmark test suite for experiments. Experimental results demonstrate that the overall performance of the MMBSO-ZSRL is significantly better than that of compared algorithms. The proposed MMBSO-ZSRL can not only find the Pareto front with better diversity and approximation, but also find more Pareto optimal solutions in the decision space. [ABSTRACT FROM AUTHOR]

The combination processing of molds and electronic product joint inspection introduce the mandatory concurrent operations on the same machine (MCDSM), which breaks the constraint that the same machine cannot process different jobs at the same time in the job shop scheduling (JSP) To solve this type of job shop scheduling problem, this paper proposed an adaptive hybrid initialization genetic algorithm (AHIGA). Firstly, this paper defined the problem as a generalized JSP considering mandatory concurrent operations on the same machine (GJSPMCO) problem, and established an optimization model using mixed integer programming with the objective of minimizing the maximum completion time. Next, it designed corresponding encoding and decoding techniques to support the representation and resolution of feasible scheduling plans within the constraints of MCOSM. Meanwhile, it established a population hybrid initialization method to generate high-quality feasible solutions. This paper also designed novel crossover and mutation operation methods to ensure the feasibility of newly generated solutions under the constraint of MCOSM. Additionally, it constructed an adaptive crossover and mutation operators to enable the adaptive updating of offspring, thereby enhancing the algorithms global search capability. Finally, it constructed 40 test cases based on JSP benchmark, and used these test cases and an example of joint inspection of electronic products for experiments. The results show that the proposed model and algorithm can effectively solve GJSPMCO problem, and the improvement strategies can enhance the quality of solutions, demonstrating the feasibility of the model and the superiority of AHIGA. [ABSTRACT FROM AUTHOR]

In order to improve the delivery efficiency of logistics, this paper proposed a vehicle routing problem with drone and time window considered by the time window, UAV (unmanned aerial vehicle) power change and multi-point continuous delivery of UAV. Then, this paper proposed a hybrid particle swarm optimization algorithm with local search to solve it. Based on the hybrid particle swarm optimization algorithm, the proposed algorithm transformed the problem solution space into the algorithm search space by constructing an efficient encoding and decoding strategy. Further, the proposed algorithm combined a single point insertion strategy, a vehicle replacement strategy and a UAV replacement strategy to form a local search strategy to improve the optimization ability. The experimental results show that the proposed model is more efficient than the model of pure vehicle distribution and saves 31.51% of the cost. The proposed algorithm is better than the four comparison algorithms, and its highest optimization rate is 82.08%. [ABSTRACT FROM AUTHOR]

Periodic high utility sequential pattern mining (PHUSPM) has attracted much attention because it can find more practical regular patterns in time series. However, existing PHUSPM algorithms struggle to effectively handle incremental updates and overlook the downward closure property and complexity of the algorithm in large-scale data. To solve this problem, this paper proposed an IncPUS-Miner algorithm, which effectively realized the incremental mining of periodic high-utility sequential patterns (PHUSPs). IncPUS-Miner introduced a novel data structure called pu-tree. Each tree node corresponded to an updated utility list(UUL) to store the auxiliary information of the corresponding sequence. When incremental data was added, this structure allowed flexible updates to project information, thereby enhancing the dynamic adaptability and scalability of the algorithm. In addition, this paper proposed two new upper bounds of sequence utility, PUB and EUB, and two corresponding pruning strategies, which effectively reduced the computational burden. The experimental results show that the IncPUS-Miner algorithm effectively realizes the incremental mining of PHUSPs on real data, and shows superior performance compared with other algorithms. [ABSTRACT FROM AUTHOR]

In order to address the limitations of the neighborhood rough set model caused by neighborhood parameters and inadequate sample information representation, this paper proposed a rough set model based on the maximum coalition theory. By normalizing the neighborhood information system and utilizing the maximum coalition set to describe neighborhood granular information, the proposed model enabled finer information division and significant reduction of uncertainty in the boundary region. Combining the model with three-way clustering, this paper designed a three-way clustering algorithm based on maximum coalition rough set. Comparative experiments on six UCI public data sets show that the proposed algorithm has better clustering quality than the comparison algorithms, and has higher comparison accuracy when dealing with boundary region samples. [ABSTRACT FROM AUTHOR]

In recent years, blockchain has attracted increasing attention in both academic and industrial circles, and the privacy of blockchain remains a controversial issue when deploying blockchain in different applications. This paper systematically reviewed the current privacy protection schemes and mechanisms in blockchain, and gave insights into the security and privacy of blockchain. Firstly, this paper briefly introduced the main functions, types and privacy definitions of blockchain technology, and analyzed the privacy issues it faced. Then, from a technical perspective, this paper divided blockchain privacy protection technology into four categories: decentralized identity authentication, privacy protocol, encryption technology and obfuscation technology, as well as eight sub-categories: decentralized identity authentication, secure multi-party computation, differential privacy, homomorphic encryption, zero-knowledge proof, coin mixing technology, ring signature and anonymous communication. This paper studied them separately, and summarized the mechanisms of different privacy protection technologies on permissioned and permissionless blockchains. Finally, this paper conducted qualitative research on the main characteristics and attributes of the eight privacy protection technologies, and discussed future research directions of privacy protection technologies. The research provided reference for blockchain developers to choose appropriate privacy protection schemes and for researchers to choose research directions for blockchain privacy protection. [ABSTRACT FROM AUTHOR]

vIn the few-shot classification tasks, existing CNN models suffer from insufficient feature extraction, limited feature diversity and weak differentiation between classes in few-shot datasets, leading to low classification accuracy. To address these issues, this paper proposed a few-shot classification model called FMAF. Firstly, this method incorporated multi-granularity thought into the architecture of CNN feature extraction network to enhance feature diversity. Secondly, after the multi-granular feature extraction network, FMAF added a self-attention layer to extract key features from the multi-granular image features, based on the multi-granular attention features, FMAF employed a feature fusion method to combine the information from multiple-granularity attention features, highlighted the crucial features and improved feature representativeness. Finally, this paper utilized two classical few-shot datasets for experimental verification on miniImageNet and tieredImageNet. Experimental results show that FMAF method can effectively improve the accuracy and efficiency of classification. [ABSTRACT FROM AUTHOR]

Most images in real life have multi-label attributes. For multi-label images, ideally, the retrieved images should be ranked in descending order of similarity to the query image, namely their numbers of labels shared with the query image decrease sequentially. However, most hashing algorithms are mainly designed for the single label image retrieval, and the existing deep supervised hashing algorithms for multi-label image retrieval ignore the ranking performance of hash codes and do not fully utilize the label category information. To solve this problem, this paper proposed a deep supervised hashing with performanceaware ranking method(PRDH), which could effectively perceive and optimize the performance of the model and improve the effect of the multi-label image retrieval. In the hash learning part, this paper designed a ranking optimization loss function to improve the ranking performance of hash codes. At the same time, this paper adopted a spatial partition loss function to divide images with different numbers of shared labels into corresponding Hamming spaces. In order to fully utilize label information, this paper also explicitly proposed using predictive label for Hamming distance calculation in the retrieval stage, and designed a loss function for multi-label classification to achieve supervision and optimization of Hamming distance ranking. A large number of results of the retrieval experiments conducted in three multi-label benchmark datasets show that the evaluation metrics of PRDH outperform the state-of-the-art hashing approaches. [ABSTRACT FROM AUTHOR]

To solve the problem that the edge devices deployed by different operators have huge investment and the cached contents are isolated and have difficulty in sharing information, this paper improved the architecture of an edge cache system based on consortium chain, so that the edge devices deployed by different operators could break the content isolation and achieve a wider range of content sharing. In order to improve the operator cache revenue while ensuring the user transmission quality and reducing the user transmission delay, this paper firstly analyzed the impact of different content delivery methods on user transmission delay and operator caching benefits in terms of cache popularity, content size, and edge device collaboration capabilities. Then, it established an optimization problem with the goal of maximizing the cache benefit and minimizing the transmission delay. Finally, to solve the proposed high-dimensional large-scale caching decision problem, it used asynchronous advantage actor-critic (A3C) based content caching algorithm to determine optimal placement of content. Simulation results show that the proposed cache strategy can effectively enhance the revenue of edge caching, reduce the delay of content delivery and improve the user experience. [ABSTRACT FROM AUTHOR]

Backpropagation-based training methods for SNNs still face many problems and challenges, including that the spike firing process is non-differentiable and spike neurons have complex spatiotemporal dynamics processes. In addition, SNNs backpropagation training methods often do not consider the relationship of the error signal between adjacent spikes, greatly reducing the accuracy of the model. To this end, this paper proposed a cross-spike error backpropagation training method for deep spiking neural networks (CSBP), which divided the error backpropagation of neurons into two dependencies: the dependency of spike firing time with the postsynaptic membrane potential (DSFT) and the dependency between spike firing time (DBSFT). Among them, DSFT solved the problem of spike non-differentiability and DBSFT clarified the dependence between spikes, allowing error signals to propagate across spikes, improving biological rationality. In addition, this paper solved the problem of insufficient expressive ability in early spiking ResNet network architecture by modifying the structural order of the spike residual block. Experimental results show that the proposed method is significantly improved compared to the SOTA (state-of-the-art) training algorithms based on spike time. Under the same architecture, the improvement is 2.98% on the CIFAR10 dataset, and 2.26% on the DVS-CIFAR10 dataset. [ABSTRACT FROM AUTHOR]

SaaS is a cloud service model where users obtain software access rights by paying subscription fees. Due to the diversity of business operations, users exhibit significant variations in the online access rates for different software. Consequently, there are variations in the cloud computing resources consumed by different software applications. To avoid the risk of violating SLAs and incurring penalty payments, SaaS operators optimize the computational resource allocation for various software applications and impose subscription limits on each category of software. Considering SLA constraints, this paper formulated a resourceconstrained nonlinear integer programming model with the objective of maximizing revenue. Due to the computational complexity of the model, it cannot be solved in polynomial time, and this paper proposed a Q-learning-PSO hybrid algorithm for this NP-hand problem. This algorithm embedded Q-learning into PSO to dynamically adjust PSO parameters, thereby avoiding the issues of local optima and low computational efficiency associated with direct PSO application. Simulation experiments validate the effectiveness of the model and algorithm in different scenarios. The results indicate that the algorithm can achieve higher revenue for subscription limits and resource allocation with superior solving efficiency under the condition of limited cloud computing resources. Specifically, in scenarios with significant demand fluctuations, operators should aim to reduce the resource contention ratio of software. This can be achieved by provisioning an ample amount of virtual machine resources and enforcing strict subscription limits to ensure the quality of service, consequently reducing penalty payments. Conversely, in scenarios with minimal demand fluctuations, operators have the flexibility to increase the resource contention ratio of software. By relaxing subscription limits, they can seize a larger market share, thus realizing revenue maximization. [ABSTRACT FROM AUTHOR]

Predictive process monitoring can provide timely information during the operation of business processes, in order to take measures to address potential risks. How to improve the accuracy of process prediction has always been highly concerned. Most of the existing research methods focus on process prediction in static environments, with few combining digital twin technology for process prediction in dynamic environments. To this end, this paper proposed a method based on concept drift detection and constructed a digital twin process prediction model to predict the next activity. Firstly, this method used behavioral relationship between event streams and weight divergence to extract features from activities in the process and obtained the feature sets of data flows. Secondly, this method performed drift detection. It dynamically selected feature sets and input them into the artificial intelligence model for training and predicting the next activity. Then, it used advanced technologies such as the Internet of Things and cloud computing to create a digital twin virtual environment. Finally, this paper obtained a digital twin model based on concept drift. It carried out evaluation and analysis on publicly available datasets, and the experimental results show that the proposed method can improve the effectiveness of prediction. [ABSTRACT FROM AUTHOR]

In the airport area, the charging behavior of new energy special vehicles has great randomness, and the charging conditions of different types of special vehicles are different, which results in a large difference in the utilization rate of each charging pile in the airport area, affecting the healthy operation of electric distribution network of the airport. In response to the above phenomenon, this paper designed a two-stage collaborative charging scheduling strategy for eleven types of special vehicles. In the first stage, this paper analyzed the service process of different vehicles for flights, with the goal of minimizing the difference value in the start time of service for adjacent flights by the same vehicle, and generated a sequence of vehicles with charging needs. In the second stage, the strategy aimed to reduce the variance of time utilization of charging piles and vehicle charging queue time in each zone of the airport area. Based on sequence of vehicles in the previous stage, this paper used an improved adaptive mutation particle swarm optimization algorithm to solve this model, and validated the strategy by comparing it with actual vehicle charging data of a domestic hub airport. The experiment shows that this algorithm can achieve a 93.5% reduction in the waiting time for vehicle charging, and it reduces the overall variance of time utilization of charging piles in the airport area by 88.7%. Finally, this strategy achieves the goal of balanced use of charging piles. [ABSTRACT FROM AUTHOR]

Existing research on knowledge base question answering (KBQA) typically relies on comprehensive knowledge bases, but often overlooks the critical issue of knowledge graph sparsity in practical applications. To address this shortfall, this paper introduced a knowledge representation learning method that transforms knowledge bases into low-dimensional vectors. This transformation effectively eliminated the dependence on subgraph search spaces inherent in traditional models and achieved inference of implicit relationships, which previous research had not explored. Furthermore, to counter the propagation of errors in downstream question-answering inference caused by semantic understanding errors of questions in traditional KBQA information retrieval, this paper introduced an answer inference re-ranking mechanism based on knowledge representation learning. This mechanism utilized pseudo-twin networks to represent knowledge triplets and questions separately, and integrated features from the core entity attention evaluation stage of upstream tasks to effectively re-rank the answer inference result triplets. Finally, to validate the effectiveness of the proposed algorithm, this paper conducted comparative experiments on the China Mobile RPA knowledge graph question-answering system and an English open-source dataset. Experimental results demonstrate that, compared to existing models in the same field, the proposed method performs better in multiple key evaluation indicators such as hits@n, accuracy, and F1-scores, proving the superiority of the proposed KBQA answer inference reranking algorithm based on knowledge representation learning in handling implicit relationship inference in sparse knowledge graphs and KBQA answer inference. [ABSTRACT FROM AUTHOR]

To address the issues of existing misinformation detection approaches, which primarily focus on single-modal data analysis and ignore the correlation between information during detection, this paper proposed a multimodal misinformation detection model combined with the social network graph (MMD-SNG model). This model used the pre-trained Transformer model and the image caption model to extract the semantics of each modality from multiple perspectives. It incorporated the features of propagated information into the text and image data by fusing the social network graph of the information dissemination process. Finally, this model used a multimodal co-attention mechanism to allocate the weights of each modality for misinformation detection. This paper conducted comparative experiments on two real datasets including Twitter and Weibo, the proposed MMD-SNG model achieved a consistent detection accuracy of approximately 88%, which was higher than existing misinformation detection approaches such as EANN and PTCA. The experimental results demonstrate that the proposed model can fuse multimodal information effectively to improve the accuracy of misinformation detection. [ABSTRACT FROM AUTHOR]

Aiming at the problem that the credibility of edge nodes in the Internet of Vehicles could not be guaranteed, this paper proposed a reputation-based task offloading and resource allocation model for the Internet of Vehicles, and used the reputation of edge nodes recorded on the blockchain to evaluate its credibility, so as to help the terminal devices select reliable edge nodes for task offloading. At the same time, this paper modeled the offloading strategy as the time delay and energy consumption minimization problem under the reputation constraint, and used the multi-agent deep deterministic policy gradient algorithm to solve the approximate optimal solution of the NP-hard problem. The edge server received rewards based on the completion of task offloading, and then updated the reputation recorded on the blockchain. Simulation experiments show that the proposed algorithm reduces in terms of time delay and energy consumption by 25.58% to 27.44% compared with the benchmark testing schemes. [ABSTRACT FROM AUTHOR]

Most traditional drone cross-domain authentication schemes rely on PKI, which are highly dependent on trusted third parties and prone to single points of failure and low authentication efficiency. Accordingly, this paper proposed a block-chain-based UAV network cross-domain identity authentication protocol scheme, which not only effectively realized the cross-domain authentication of UAVs, but also improved many security problems faced by the traditional PKI-based model. At the same time, to discover the privacy protection of UAV network cross-domain authentication, this paper proposed a method to anonymize the UAV cross-domain identity authentication identity. The results of security analysis and simulation experiments show that the scheme can not only resist common network attacks, but also significantly improve authentication efficiency and reduce time overhead. [ABSTRACT FROM AUTHOR]

The digital trunking system has the characteristics that the calling mode is mainly group calling and the communication mode is mostly half duplex. For the problem that the traditional user classification method couldn't meet the classification requirements of digital cluster users, this paper proposed a digital cluster user classification method based on improved RFM model. Firstly, it introduced the average speech duration attribute to establish the recency vitality speak (RVS) model. Then, it used the analytic hierarchy process to determine the weight of each parameter in the model. Finally, it used the K-means + + clustering algorithm to classify digital cluster users. The simulation result shows that, by using the user classification method proposed in this paper, the accuracy of digital cluster user classification can reach more than 87. 9% . [ABSTRACT FROM AUTHOR]

For the security problem of the spread of digital image and the digital image encryption from plaintext and too depend on the chaotic systems, this paper proposed a digital image encryption and monitoring algorithm combined with plaintext-related chaotic mapping and SHA-256. The algorithm used the SHA-256 algorithm to calculate the hash value as record to monitor the spread of digital image. Lorenz chaotic mapping generated the corresponding cryptogram by using forward diffusion, associated plaintext scrambling and backward diffusion to encrypt the digital image. The results show that the algorithm can resist many attacks and achieve the security and concealment of image propagation. [ABSTRACT FROM AUTHOR]

In recent years, the number of papers in scientific social network had grown at an explosive rate. It was difficult for users to find papers related to their requirement. And the paper recommendation is one of the key methods to solve this problem. However, most of the existing methods only focus on rating prediction, ignoring the ranking problem between papers. In addition, a lot of social information in scientific social network were not fully considered in the traditional recommendation method. Therefore, this paper proposed an improved List-wise method for scientific paper recommendation. It systematically analyzed The friendship information between users , the titles , abstracts and tags of scientific papers and incorporated theses information into the improved List-wise method. In order to verify the validity of the proposed method , this paper crawled data from a scientific social network , i. e. , CiteULike , to conduct experiments. Experimental results show that the proposed method gets the best recommendation results and performs well in scalability compared to the other traditional recommendation methods. [ABSTRACT FROM AUTHOR]

Paper folding is a common phenomenon. This paper could be referred to as the shell, and the model was the key for efficient simulation of thin shell. Based on the high cost and more complex of physical modeling and based on the accuracy of deformation was not enough of geometric modeling. Therefore, in order to realize the simulation of paper folding, this paper proposed a paper folding simulation based on physical model and geometric model. First of all, based on the hinge model and the mass spring model of paper thin shell model. Second, updating the position of the particle by the implicit integration method, it determined the initial deformation of the paper,and increased the deformation constraint. Lastly, according to Poisson deformation determining the final deformation state. The experimental results show that the hybrid model has the advantages of high speed and high stability. [ABSTRACT FROM AUTHOR]

Aiming at the situation of imperfect spectrum sensing of cognitive radio in actual cognitive ultra-dense network sce- narios, this paper proposed a resource allocation scheme based on imperfect spectrum sensing. The target was to maximize the energy efficiency of secondary network in cognitive ultra-dense network with imperfect spectrum sensing while taking into ac- count cross-tier/co-tier interference constraints and ensuring users quality of service. And this paper constructed the energy efficiency optimization problem based on the network model, which was a mixed integer non-convex programming problem, then transformed it into an equivalent convex optimization problem by the time-sharing relaxation method and Dinkelbach me- thod, and finally used the Lagrangian dual method to get its optimal solution, thereby obtained the sub-channel and power allo- cation strategy for optimal energy efficiency. Based on this, this paper proposed an iterative sub-channel and power allocation algorithm. In order to trade off the computational complexity, this paper also proposed a practical sub-channel and power allo- cation algorithm. Simulation results show that the proposed algorithms effectively improve energy efficiency. [ABSTRACT FROM AUTHOR]