最新刊期
卷 23 ， 期 12 ， 2024
- “在脑机接口领域，研究者提出了改进的动态图注意力通道选择方法，有效识别认知工作负荷，为复杂数据环境下的模型鲁棒性提供解决方案。”摘要：Research on the recognition of cognitive workload based on electroencephalogram （EEG） and functional near-infrared spectroscopy （fNIRS） physiological data has garnered significant attention in the field of brain-computer interfaces. However， the complex data acquisition environment introduces uncontrollable effects on inter-channel data， severely limiting the accuracy and integrity of models simulating human brain information transmission processes. Therefore， this paper proposes an improved dynamic graph attention-based channel selection method. The method utilizes attention scores returned by a Graph Attention Network （GAT） to select channels， thereby reducing environmental interference and enhancing model robustness. Moreover， simple feature fusion can overlook the heterogeneity between different modalities， leading to the loss of critical information. To mitigate this， we designed a hierarchical feature fusion module. We validated our approach using two publicly available datasets provided by the Berlin Institute of Technology： a mental arithmetic task and an N-Back task. Employing a subject-dependent training strategy and ten-fold cross-validation for each participant， our method achieved average accuracies of 85.44% and 91.72%， respectively. Compared to current state-of-the-art methods， our approach demonstrates certain advantages. The experimental results indicate that the proposed model effectively recognizes cognitive workload in complex data environments.Additionally， the proposed channel selection method is significant for reducing computational cost and eliminating irrelevant channels.关键词：electroencephalogram;functional near-infrared spectroscopy;channel selection;cognitive workload relognition;hierarchical feature fusion0|0|0发布时间：2024-12-30
- “在风电功率预测领域，专家提出了基于特征重组和优化算法的BILSTM与RF组合模型，显著提升了预测精度，R2达到0.99425。”摘要：Short term wind power prediction is crucial for the normal operation of the power system. In order to improve the accuracy of wind power prediction， a combination model of bidirectional long short-term memory network （BILSTM） and random forest （RF） is proposed based on feature recombination method and improved quantum particle swarm optimization algorithm （IQPSO） to optimize the short-term wind power prediction. Firstly， using local mean decomposition to process wind power data， multiple sub components are obtained， and their fuzzy entropy is calculated to recombine new feature components. Secondly， using IQPSO optimized BILSTM to predict feature components， the results of each component are superimposed to obtain preliminary predicted values. Finally， error correction was performed on the preliminary predicted values using IQPSO optimized RF. The experiment showed that the coefficient of determination （R2） of the model reached 0.994 25， which is superior to other models. The ablation experiment verified the necessity of each module.关键词：wind power prediction;feature recombination;improved quantum particle swarm optimization algorithm;bidirectional long short-term memory network;random forest;error correction0|0|0发布时间：2024-12-30
- “在云计算领域，专家提出了MVMD-MHAT-BiLSTM模型，优化资源负载预测，提高了预测精度。”摘要：Cloud computing service providers currently face huge challenges in predicting large-scale workloads and resource usage. Due to the difficulty in capturing nonlinear characteristics， traditional prediction methods usually cannot achieve high prediction performance for resource load data. In addition， there is a lot of noise in the original time series. If smoothing algorithms are not used to denoise these time series， the forecast results may not meet the provider's requirements. To this end， this paper proposes a MVMD-MHAT-BiLSTM combined prediction model. This model first uses the improved gray wolf optimization algorithm to optimize the VMD parameters， and then uses the variational mode decomposition signal decomposition method to decompose the complex， nonlinear original The time series is decomposed into low-frequency intrinsic mode functions； then a multi-head attention mechanism is introduced in BiLSTM to capture multi-level， bidirectional features； and finally the attention mechanism is used to explore the importance of different output dimensions. Taking the CPU usage of machines in Alibaba Cloud Cluster Data as an example，compared with theBiLSTM，Pa-BiLSTM，CNN-BiLSTM，MHAT-BiLSTM and VMD-MHAT-BiLSTM， the RMSE of of the model proposed in this article decreased by 8.6% to 19.3%， achieving higher prediction accuracy.关键词：cloud resource load prediction;GWO;VMD;multi-head attention mechanism;BiLSTM;attention mechanism0|0|0发布时间：2024-12-30
- “在交互式推荐系统领域，专家提出了基于对比学习和深度强化学习的推荐系统框架，有效解决了正负反馈建模问题，为推荐系统研究开辟了新方向。”摘要：The application of deep reinforcement learning techniques in interactive recommendation systems has reached a high level of maturity. However， there is currently limited research dedicated to modeling there presentation of states. Existing works primarily focus on modeling state representations based on positive feedback sequences during user interactions. This approach results in the oversight of potential relationships existing within negative feedback sequences generated by users during interactions， as well as changes in user interests. Consequently， the recommendations produced by such systems tend to be one-sided. To address this gap， a novel recommendation system framework， named Contrastive Learning and Deep Reinforcement Learning-Based Recommender System （CRLRS）， is proposed. CRLRS is designed to model state representations for both positive and negative feedback sequences generated during user interactions. Additionally， in order to mitigate data sparsity issues associated with positive feedback and address differences between fine-grained positive and negative feedback， a contrastive auxiliary task is incorporated. Extensive experiments were conducted on two real-world datasets， among which HR@10 The results of the evaluation indicators on the Movielens-100k and Movielens-1m datasets are 0.705 2 and 0.490 2， respectively； NDCG@10 The results of the evaluation indicators are 0.478 2 and 0.271 5. The comparison results show that our method is significantly better than the current state-of-the-art methods， which proves the necessity of CRLRS modeling positive and negative feedback simultaneously and adding comparative auxiliary tasks， and has better recommendation performance.关键词：deep reinforcement learning;contrastive learning;recommender system;positive negative feedback;state representation0|0|0发布时间：2024-12-30
- “在自然语言处理领域，针对情感分析的少样本问题，研究者提出了基于ChatGLM模型的多阶段数据增强策略，有效提升了准确度。”摘要：Sentiment analysis is one of the popular tasks in natural language processing. Due to the difficulty and high cost of annotating training data， sentiment analysis with limited samples has drawn people's attention. Data augmentation methods are one of the primary approaches for handling limited sample learning. However， traditional data augmentation methods have not taken into account the characteristics of sentiment analysis， which can lead to issues such as semantic inconsistencies， sentiment bias， and excessive generation in the augmented data. To address these problems， a multi-stage data augmentation strategy based on the ChatGLM model is proposed specifically for sentiment analysis. Specifically， it starts with simple word-level data augmentation using EDA methods， followed by filtering the generated data using a sentiment lexicon， and finally， enhancing it at the sentence level using the ChatGLM model. Experimental results demonstrate that this data augmentation method improves accuracy by 1.9%， 2.1%， and 2.2% on three different datasets compared to the traditional optimal data augmentation method， confirming the effectiveness of this approach for limited sample sentiment analysis.关键词：few-shot learning;sentiment analysis;data augmentation;pre-trained models;natural language processing0|0|0发布时间：2024-12-30
- “在工业部件异常检测领域，研究人员提出了一种基于多尺度特征融合的自监督算法，有效提高了检测准确性，为工业生产质量控制提供了新方案。”摘要：Industrial component anomaly detection is a key issue in industrial production， where the main objective is to detect and identify anomalous components in time to ensure product quality and production efficiency. However， current industrial component anomaly detection algorithms are still extremely challenging， such as the impact of target defects at different scales on the accuracy of the algorithms， the uncertainty that all possible anomaly data cannot be exhausted. To solve the above problems， proposed a self-supervised anomaly detection algorithm for industrial components based on multi-scale feature fusion. Using Poisson fusion to seamlessly integrate rectangular blocks of different sizes into normal samples to generate anomaly sample label pairs， and proposes an Attention Atrous Spatial Pyramid Pooling（A-ASPP） module based on a CNN model with encoder-decoder structure， which achieves multi-scale feature extraction of images through Atrous Spatial Pyramid Pooling， and uses channel attention mechanism and spatial attention mechanism to achieve multi-scale feature interaction and focus region weights， and finally locates anomalous regions through the probability map output by the model. The experimental results show that the AUROC metric of this paper's method improves by 9.2% compared to the NSA method for the screw category in the public dataset MvTecAD. The method in this paper achieves an average AUROC of 98.5% on this dataset， superior to NSA methods.关键词：self-supervised learning;multi-scale feature fusion;Poisson integration;attention mechanism0|0|0发布时间：2024-12-30
- “在大学生情感分析领域，研究者提出了基于FastText字向量表示方法结合Self-Attention与BiLSTM的分析方法，实验显示其分类性能较传统模型提高了6%和3%。”摘要：The performance of the network model based on the word vector relies heavily on the accuracy of word segmentation，a method of sentiment analysis for college students based on FastText character vector combined with Self-Attention and BiLSTM is proposed. Firstly， character vectors are generated using the fasttext model， then contextual semantic features are extracted by the bidirectional long and short-term memory model and key information is strengthened using the Self-Attention mechanism， finally，the sentiment categories are judged using the Softmax classifier. The experimental results show that character vector is more suitable for short text than word vector，and character-SATT-BiLSTM has achieved better classification results than character-LSTM， character-BiLSTM and other models. The classification performance can be increased by 6% and 3%， respectively.关键词：FastText;character vector;bidirectional long short-term memory;self-attention;emotional tendency analysis0|0|0发布时间：2024-12-30
- “深度学习领域专家提出了基于图卷积网络的两阶段异常检测方法，有效提升了复杂网络结构的异常检测性能。”摘要：The field of deep learning is paying increasing attention to graph structured data， and multiple fields have abstracted entities into attribute networks. Knowledge graphs and other organizational methods have successfully achieved efficient organization and management of information. In these information rich networks， security issues are particularly important as the presence of anomalous entities may pose a threat to overall interests. Traditional methods face certain difficulties in anomaly detection of graph structured data， especially in capturing high-dimensional network features. Although deep learning methods are powerful， due to the limitations of network depth， it is often difficult to obtain global information from the network. Therefore， a two-stage anomaly detection method based on graph convolutional neural network is proposed， which gradually obtains the community information of nodes through graph convolutional neural network， overcoming the shortcomings of traditional methods in capturing high-dimensional features； Simultaneously considering the node's own attributes to better adapt to various complex network structures and improve anomaly detection performance. The experimental results show that the AUC score of this method exceeds 0.9 on some datasets， and it can achieve optimal or suboptimal performance compared to the baseline method on each dataset.关键词：anomaly detection;graph convolutional networks;community detection;attributed network0|0|0发布时间：2024-12-30
- “在物资招标领域，专家提出了基于K-BERT-LDA的层级多标签文本分类方法，通过特征融合和标签树形结构关系提升分类准确性，为智能分标提供解决方案。”摘要：Traditional manual bidding has low efficiency and accuracy in dividing bids. A hierarchical multi label text classification method based on K-BERT-LDA is proposed for material bidding texts with sparse semantic features and obvious hierarchical structure of labels. Firstly, text features are extracted through a hybrid model. The K-BERT model extracts text features with knowledge injection to compensate for semantic information gaps. The LDA topic model extracts topic distribution features and further enriches the text feature representation through feature fusion. Secondly, joint embedding of category labels, where the prediction results of upper level labels can guide lower level classification and fully utilize the tree structure relationship between labels to improve the accuracy of multi label text classification. Finally, an intelligent processing strategy based on text similarity algorithm is proposed to ensure the success rate of bidding and obtain bidding results by merging sections with insufficient pre investment amounts. The experiment shows that the proposed method has better classification performance than other classification methods and single model, and the accuracy, precision and F1 value reach 95.45%, 92.57% and 91.88% respectively, which can effectively and accurately achieve the goal of intelligent classification.关键词：bidding division;hierarchical multi-label text classification;knowledge injection;topic distribution;feature fusion;text similarity0|0|0发布时间：2024-12-30
- “在烟叶打叶复烤工艺领域，专家构建粒子群优化随机森林模型，探究参数对出口指标影响，为提升烟叶复烤质量提供理论依据。”摘要：The moisture content and temperature stability at the exit of the secondary moistening leaves are the key indexes to evaluate the recuring process of tobacco leaves. However， it is difficult to accurately control the outlet index of secondary moistening in a regrilling plant in Yunnan province due to parameters such as ambient temperature and water steam flow. Through the construction of random forest algorithm model based on particle swarm optimization， the influence of various parameters on the export index of two rungs under different working conditions was explored. After cleaning the historical data of secondary leaf wetting parameters， Pearson coefficient analysis was carried out after removing dirty data to find the key production control parameters closely related to export quality. Combined with field manual experience and correlation analysis， the random forest algorithm of particle swarm optimization was used to optimize the return air temperature， hot air temperature， drain damper and compensation steam valve opening， and compared with random forest， gray wolf optimization random forest and BP neural network. The results show that the mean square error of return air temperature and hot air temperature obtained by the proposed algorithm is 0.003， and the mean square error of the opening of the tidal damper and the compensating steam valve is 0.001. The algorithm provides a theoretical basis for operators to adjust the equipment and improve the quality of tobacco recuring.关键词：secondary leaf rinsing;random forest;particle swarm optimization;correlation analysis;Mean square error0|0|0发布时间：2024-12-30
-
Improved Global Pointer Based Named Entity Recognition Method for Enterprise-benefiting Policies AI导读
“最新研究突破，专家提出基于多级词汇全局指针与对抗训练的命名实体识别模型，有效提升中小微企业政策信息提取效率。”摘要：Small and medium-sized enterprises play an important role in the national economy. In recent years， various preferential policies for enterprises introduced by the government have included key information for government decision-making. However， policy texts have complex structures， strong dependence on professional semantics， and contain noisy text and nested entities， making information extraction difficult. Therefore， a named entity recognition model based on multi-level vocabulary global pointers and adversarial training is proposed. This model integrates the LEBERT model at the embedding layer to obtain the combined semantic representation of characters and vocabulary， and constructs a global entity matrix through global pointers to uniformly process flat and nested entities； Simultaneously introducing rotary position encoding to enhance the perception of position information， and combining it with adversarial training to enhance stability and robustness. The experimental results show that the F1 value of the model is 81.90%， which is 4.72% higher than the classical sequence annotation based model. The overall performance supports downstream task development.关键词：named entity recognition;enterprise-benefiting policies;pre-training model;global pointer;adversarial training0|0|0发布时间：2024-12-30
Artificial Intelligence
- 摘要：Age of Information （AoI） is a measure of data freshness， which is an important consideration in the optimization design of energy harvesting IoT. Existing AoI optimization strategies are often designed only from the source node side or the base station side. In this regard， a joint optimization scheme that takes into account both the source node sampling strategy and the base station scheduling strategy is designed for the energy harvesting IoT with multiple source nodes， an energy cap， and an ideal channel. The source node sampling strategy considers the following tradeoffs： the earlier the sampling time， the larger the AoI of the cached samples； if the sampling time is too late， the energy may be wasted or too late to be uploaded to the base station. The base station scheduling is based on a greedy strategy， i.e.， selecting the source node that minimizes the long-term average AoI increase of the system. Simulation analysis shows that the method can obtain better long term average AoI compared to existing proven baseline methods.关键词：age of information;energy harvesting IoT;sampling strategy;scheduling strategy0|0|0发布时间：2024-12-30
- “在SLAM领域，专家提出了基于环境判断的激光视觉融合定位算法，有效提高了复杂环境下的定位精度。”摘要：In order to solve the problem of low positioning accuracy of SLAM systems in certain low texture or low obstacle scenes， a laser vision fusion positioning algorithm based on environmental judgment is proposed. This algorithm compares the amount of information obtained by laser sensors and visual sensors to determine whether the current environment is favorable for laser sensors or visual sensors. Set a weight for each sensor and optimize the pose based on the weight of each sensor's information to improve positioning accuracy. In addition， due to the limited texture information obtained by laser sensors， using laser sensors for loop detection results in significant errors. To improve the accuracy of loop detection， it is dynamically determined whether to use laser sensors or visual sensors for loop detection based on environmental judgment results， and then the robot pose is globally optimized to spread out errors. The experimental results show that in complex environments， the proposed algorithm has higher accuracy compared to the original algorithm. The errors in the three sequences of mh_02/easy， V1_02/medium， and V1_03-difficult are 0.031/0.025， 0.040/0.037， and 0.036/0.033， respectively， which can fit the real trajectory well.关键词：localization;complex environments;laser sensors;vision sensors;environmental judgment;loop closure detection0|0|0发布时间：2024-12-30
- “在气象雷达领域，研究者采用扩展卡尔曼滤波算法处理梯度VAD技术，有效提升了风场反演精度。”摘要：Gradient VAD technology is proposed to overcome the significant impact of velocity ambiguity on traditional VAD techniques. However， during the use of gradient VAD technology， errors or noise in radial velocity can be amplified in the radial velocity azimuth gradient. Therefore， low-pass filtering is usually used to smooth the radial velocity azimuth gradient. However， it still has a serious impact on the accuracy of horizontal wind inversion. This article proposes to use the extended Kalman filter （EKF） algorithm to process the radial velocity gradient in the azimuth direction. The results show that theextended Kalman filter algorithm can effectively reduce the influence of small fluctuations and noise in the radial velocity gradient in the azimuth direction， thereby improving the horizontal accuracy of wind field inversion. In this article， the Xiamen SA band Doppler radar data was preprocessed twice with velocity un deblurring and velocity de deblurring， and then compared with low-pass filtering and EKF. The results showed that the variance of EKF algorithm compared with low-pass filtering algorithm was reduced by about 40% and 50% respectively in the process of velocity de deblurring and non thrust blur data processing. This indicates that the radial velocity gradient after EKF processing is more stable， thus improving the inversion accuracy of gradient VAD technology.关键词：extended Kalman filter algorithm;gradient VAD;Doppler weather radar;wind field inversion;low-pass filtering0|0|0发布时间：2024-12-30
- “服务计算技术发展迅速，专家提出群组融合方法，为个性化服务定制需求提供解决方案。”摘要：With the booming development of service computing technology， the scope of services has expanded from online services to offline travel， shopping， catering and other industry sectors， generating massive demand for personalized service customization. However， given the cost of customization and other factors， service providers often do not provide personalized services for small-scale users one by one. Finding commonalities in the personalized service customization needs of a large number of users， and clustering and fusing similar needs into groups to form larger-scale group customization needs are expected to establish a win-win situation for both supply and demand. This demand grouping operation needs to be carried out based on the asymmetric similarity between demands， while existing clustering algorithms rely on similarity and do not consider the compatibility of objects after clustering. To this end， a group fusion method is proposed for personal service customization requirements， with the added constraint of satisfaction， under which the clustering of demand objects is carried out， and the calculation method of asymmetric similarity between requirements is given based on the establishment of the customization requirement model， and then the group construction and fusion algorithms are designed with the optimization objective of maximum average group satisfaction， in order to group several similar personal customization requirements into a single group and to fused into a group customization requirement with full group satisfaction， and finally demonstrated the feasibility and effectiveness of the method through specific experiments.关键词：service calculation;personalization;asymmetric similarity;group fusion;clustering;group satisfaction0|0|0发布时间：2024-12-30
- “在移动机器人路径规划领域，研究者提出了一种改进SARSA算法，有效提升了机器人智能化行动的路径规划能力。”摘要：Aiming at the problems of slow convergence speed， high exploration randomness， and poor path planning performance in mobile robot path planning using traditional SARSA algorithm， this paper proposes an improved SARSA algorithm that combines artificial potential field method with traditional SARSA algorithm. Firstly， the reward function of the algorithm is controlled by the gravity function of the artificial potential field to increase the guidance during exploration； Secondly， using the repulsive function of the artificial potential field to generate μ To effectively adjust the Q value in path planning， the value should be adjusted to lower the Q value as it approaches obstacles， thereby improving the convergence speed of the path planning algorithm and reducing the frequency of collisions with obstacles. The improved SARSA algorithm was compared with other algorithms through simulation experiments， and the results showed that the improved SARSA algorithm has significantly improved performance in convergence speed， average learning time， average learning steps， and average number of collisions with obstacles， which can effectively enhance the path planning ability of intelligent actions of mobile robots.关键词：reinforcement learning;improved SARSA;artificial potential field method;path planning;mobile robot0|0|0发布时间：2024-12-30
Computer Software and Theory
- “在未知搜救场景下，专家结合数字孪生与机器人集群技术，建立了智能搜救数字孪生框架，为高效完成搜救任务提供解决方案。”摘要：In response to the challenges of high-risk factors， inadequate global information understanding， and low decision-making efficiency in artificial rescue operations within unknown search and rescue scenarios， this study proposes integrating digital twins and robot clusters for efficient decision-making. Digital twins’ visual integration and virtual inference capabilities are leveraged for unmanned execution of search and rescue tasks. A hierarchical multi-machine intelligent search and rescue digital twin framework is proposed along with a real-time model construction method based on SLAM. Each component is introduced in detail and theoretically analyzed. To address real-time mission requirements， the study designs online synchronization modules， collaborative target search and positioning modules， and parallel decision-making rescue path planning modules. These elements aim to overcome challenges such as asynchronous dynamics and inadequate accuracy in target search and positioning. Implemented in a Unity3D-based prototype system， the framework and modules are validated in simulated scenarios， demonstrating reliability and efficiency. This research contributes valuable insights for enhancing artificial rescue operations.关键词：urban search and rescue;digital twin;parallel intelligence;unmanned systems;cluster collaboration0|0|0发布时间：2024-12-30
- “在激光扫描共聚焦显微镜领域，专家设计了基于FPGA的高速数据采集系统，实现了250 Mbps采样速率和14 bit数据采样精度，满足成像仪器需求。”摘要：Aiming at the requirement of large field of view and high resolution of laser scanning confocal microscope， a high-speed data acquisition system based on FPGA is designed in order to obtain more data in the scanning time of galvanometer. This system uses Xilinx A7 series FPGA as the main control chip， LMK0482X series clock generator provided by TI company and a single channel 14-bit high-speed ADC14X250 chip supporting JESD204B protocol to form the signal acquisition system. The system uses Verilog hardware description language to realize each module function under Vivado platform， and completes the register configuration of LMK0482X and ADC14X250 chips through SPI protocol， and constructs high-speed serial data link based on AXI protocol by using Xilinx's JESD204B IP core， realizing efficient data acquisition and transmission between ADC and FPGA. The experimental results show that the interface can receive the analytical data correctly at the transmission rate of 5 Gbps， the sampling rate of the system can reach 250 Mbps， and the data sampling accuracy is 14 bit， which meets the data acquisition requirements of laser scanning confocal imaging instruments.关键词：FPGA;high speed ADC;data acquisition;confocal microscopy0|0|0发布时间：2024-12-30
- “在医疗应急预案领域，专家构建了垂直领域知识图谱，提出开源大模型增强问答方法，逻辑准确率达84.16%，为智能化提供新方向。”摘要：At present， the medical emergency response plan field in China is large in scale but not highly intelligent. The knowledge graph question answering system can transfer the operation core from humans to machines， thus achieving intelligence. However， the high-quality knowledge graph scale in this field is relatively small， and the question answering matching task has problems of inefficient retrieval and complex processing flow. The big language model provides a new direction for knowledge graph question answering. Explore the integration of open-source big models with intelligent medical emergency systems， construct a knowledge graph in the vertical field of medical emergency， and propose a method for enhancing knowledge graph Q&A with open-source big models. This method involves post production retrieval， utilizing fine tuned open-source models to generate query statements. A dictionary composed of knowledge graph entities and relationships replaces the generated query statements with entities and relationships， and obtains knowledge graph answers through standardized query statements. After experimental testing， the logical accuracy of this method on the test set reached 84.16%， which is feasible on self built knowledge graphs and has reference value for other fields.关键词：large language model;knowledge graph Q&A;intelligence;emergency plan0|0|0发布时间：2024-12-30
Software Design, Development, and Application
- “在企业信用管理领域，专家基于区块链提出了多级征信数据安全存储和共享系统，有效解决了数据孤岛、单点故障与隐私泄露等问题，为数据安全存储和可靠共享提供解决方案。”摘要：In the current enterprise credit management system， data held by various entities is both diverse and heterogeneous， leading to significant challenges in data sharing. Compounding this scenario are issues such as data silos， single points of failure， and privacy breaches. Blockchain technology， renowned for its inherent decentralization and tamper-proof characteristics， effectively addresses these challenges. Accordingly， this paper introduces a multi-tiered credit data storage and sharing system based on blockchain technology. Initially， the system employs a combination of on-chain and off-chain storage to achieve multi-tiered， hybrid data storage， thus overcoming the storage bottlenecks associated with blockchain. Subsequently， a priority-based， multi-level priority queue sorting algorithm is proposed. This innovative algorithm ensures equitable resource allocation and performance-prioritized scheduling for multi-level data， thereby enhancing the overall quality of blockchain services. Furthermore， a fine-grained access control strategy is developed. Integrating both channel and file encryption， this strategy effectively mitigates potential file leakage issues in the InterPlanetary File System （IPFS） and substantially improves the security and reliability of blockchain file storage. Experimental results indicate that the system not only effectively resolves the redundancy and security issues inherent in traditional sharing models but also achieves efficient， secure data storage and reliable sharing， all while fulfilling the performance demands of practical applications.关键词：blockchain;data sharing;access control;data storage0|0|0发布时间：2024-12-30
- “在在线招聘领域，专家提出了基于分布式数字身份的隐私保护方案，有效解决了用户隐私泄露和招聘信息伪造问题，为在线招聘业务隐私保护提供了新思路。”摘要：Aiming at the problems of user privacy leakage and forged recruitment information in the current business scenario of online recruitment， a distributed digital identity-based online recruitment privacy protection scheme is proposed by introducing distributed digital identity and public key encryption technology. Through the infrastructure of distributed digital identity， the privacy data is returned to the users and no longer stored in the three-party database， and the proxy re-encryption technology supports the secure flow and verification of data. For each process in the scheme， algorithms were designed to test the performance under high concurrency respectively. The experimental results show that digital credentials based on distributed digital identity can provide efficient and low-cost credential issuance and authentication services， and protect the privacy security and effective authentication during data flow with acceptable performance.关键词：distributed digital identity;online recruitment;digital certificate;proxy re-encryption0|0|0发布时间：2024-12-30
-
Design and Implementation of Cold Fresh Beef Traceability System Based on Blockchain Technology AI导读
“区块链技术助力牛肉冷链溯源，提升数据透明度和追溯效率，为智能保质储运提供技术支持。”摘要：Beef cold chain traceability has difficulties such as difficult detection， centralized information， long traceability time， non-transparent meat information， and non-uniform verification methods. The use of blockchain ledger decentralization， non-tampering， data transparency and other characteristics can solve the above problems. Cold chain data is collected through IoT devices， and blockchain CA is used to realize device authentication and ensure reliable source data. Based on Hyperledger Fabric， it builds a coalition chain network with supervisory nodes， and utilizes smart contracts to complete data verification and chain code supervision. At the same time， by adopting a variety of storage technologies， the read/write efficiency of the blockchain network is effectively improved. Taking the cold beef cold chain as an example， the blockchain parameters are tuned and tested under the premise of meeting the resource utilization requirements. The results show that the average transaction throughput of the whole node network is 154 TPS （tx/s）， and the average maximum transaction latency is 0.14 s. The average transaction throughput of the subchain reaches 270 TPS， the average maximum transaction latency is 0.084 s， and the concurrent processing transaction throughput ranges from 200 TPS to 500 TPS. While meeting the traceability requirements of the traceability system， it provides technical support for intelligent quality preservation storage and transportation of chilled beef.关键词：blockchain;Hyperledger Fabric;traceability;smart contract;cold chain0|0|0发布时间：2024-12-30 - “在智能电网领域，专家提出了一种基于区块链的可追溯匿名认证方案，有效解决了低延迟和实时同步信息的挑战，为智能电网安全高效运行提供了新思路。”摘要：Low latency and real-time synchronization of information are currently two challenges faced by smart grids. Most existing smart grid identity authentication schemes suffer from high computational and communication costs， making them unsuitable for smart devices with limited computing resources in smart grids. A traceable anonymous authentication scheme for smart grids based on blockchain is proposed. This scheme can achieve edge server authentication of legitimate smart meters under the condition of anonymity of smart meters. At the same time， the authentication process does not require the participation of the power service registration authorization center. In order to solve the efficiency problem， this scheme constructs a chameleon hash algorithm that can significantly improve the efficiency of subsequent authentication after the fist authentication. Security analysis shows that this scheme has indistinguishability， anonymity， conditional traceability， and can resist simulation attacks and replay attacks. Performance analysis shows that this scheme can meet the requirements of low latency and efficiency in smart grid environments.关键词：identity authentication;blockchain;anonymity;condition traceability;smart grid0|0|0发布时间：2024-12-30
Blockchain Technology and Applications
- “三维网格简化技术取得新进展，改进QEM算法有效保留特征点，提升视觉效果连续性。”摘要：3D mesh simplification is an important technology in computer graphics， widely used in fields such as virtual reality， game development， and computer animation. Although traditional QEM algorithms can effectively reduce the complexity of models during simplification， they have certain limitations in preserving important details and features. To solve this problem， the QEM algorithm introduces the determination of the angle between the normal vectors of the triangle before and after folding， as well as the regularity of the triangle as a constraint condition for edge folding. This improvement strategy aims to ensure the continuity of the visual effects of the model during the simplification process and minimize the generation of narrow triangles. Experiments were conducted on multiple 3D models， and the results showed that the improved algorithm is more effective in preserving feature points compared to traditional QEM algorithms and vertex clustering algorithms under the same simplification rate conditions， ensuring similar visual effects before and after simplification. Additionally， the improved algorithm minimizes the Hausdorff distance between the processed model mesh and the original model mesh.关键词：mesh simplification;secondary error measurement;normal vector;triangle regularity;edge folding0|0|0发布时间：2024-12-30
- “在遥感图像道路提取领域，MSS-LinkNet模型通过多尺度卷积注意力模块和条形池化模块，有效提升了道路检测的准确性和效率。”摘要：In the task of extracting roads from remote sensing images， road information is often affected by environmental factors such as lighting， shadows， and occlusion， and roads usually appear as slender strips， making it difficult to accurately detect. To this end， an improved LinkNet model （MSS LinkNet） for multi-scale and strip features is proposed to capture contextual information at different scales， which is highly compatible with the slender characteristics of roads. Firstly， the multi-scale convolutional attention module is used as the basic component unit of the encoder to ensure the model's ability to extract multi-scale and stripe features. Secondly， an improved hollow space pyramid pooling module is added to the central area of the network to enhance the model's ability to parse multi-scale information. Finally， a bar pooling module is added to the decoder section to enhance the model's ability to parse bar information. The experiment shows that compared to D-LinkNet， the proposed model has improved IOU by 2.53% and 0.71% on the DeepGlobe and Massachusetts datasets， respectively， while only accounting for 54.15% and 79.63% of D-LinkNet in terms of parameter and computational complexity.关键词：road extraction;multi-scale feature;strip feature;attention mechanism0|0|0发布时间：2024-12-30
- “在滇中高原农作物种植结构提取领域，专家基于Sentinel-2A影像数据，构建面向对象的支持向量机模型，总体精度达93.87%，为当地农业灌溉和经济发展提供帮助。”摘要：The high altitude of the Central Yunnan Plateau region results in fragmented crop distribution and small planting areas. Obtaining the crop planting structure quickly and accurately is of great significance for local agricultural irrigation and yield estimation. At present， there is little research on the complex crop areas in the central Yunnan Plateau based on Sentinel-2A image data. Therefore， a neural network， support vector machine， and random forest classifier are constructed based on the combination of spectral， texture， and terrain features. The suitable feature combination and optimal classifier for irrigation areas are analyzed and compared. The experimental results show that among the three classification models， support vector machines are more suitable for extracting planting structures in irrigation areas， with an overall accuracy of 91.74% and a Kappa coefficient of 0.87. On this basis， an object-oriented support vector machine model was constructed， and the overall accuracy of crop extraction was further improved， with an overall accuracy of 93.87% and a Kappa coefficient of 0.90. Compared with the traditional three feature combination support vector machine method， the overall accuracy was improved by 2.13%. The object-oriented support vector machine method is suitable for crop classification in the large-scale irrigation area of Qingling River in the central Yunnan Plateau， and can provide assistance for local water conservancy irrigation and agricultural economic development.关键词：Sentinel-2A;planting structure;feature extraction;support vector machine0|0|0发布时间：2024-12-30
- “在公路路面裂缝检测领域，研究者提出了基于改进Yolov5s的病害检测模型，有效提升了检测精度，为路面养护提供新方案。”摘要：Highway pavement cracks are an important influencing factor in asphalt pavement diseases， and pavement crack detection is an important part of pavement maintenance. A pavement crack disease detection model based on improved Yolov5s is proposed to address the problems of missed detections， false detections， and low recognition accuracy in detection algorithms for highway pavement cracks. Firstly， the attention mechanism module CBAM is adopted to learn target features and positional features， and to increase useful feature weights； Secondly， the Decoupled decoupling head method is proposed to separate the feature maps through different branches for processing， in order to improve training accuracy； Finally， an improved α DIoU loss function is proposed to replace the CIoU loss function in the original model， and α=3 is selected to enhance the loss gradient value of the high IoU object and the regression effect of the box. The experiment shows that the improved model has an average detection accuracy of 92.8%， a recall rate of 94.5%， and an mAP value of 96.5%， which is 1.8% higher than the original model. This proves that the improved model has a high improvement effect on detection accuracy and can meet the recognition and detection tasks of highway pavement cracks.关键词：road crack detection;Yolov5s;attention mechanism;decoupling head;loss function0|0|0发布时间：2024-12-30
- “在草坪环境下行人检测领域，YOLO-CGO模型通过轻量化改进，显著提升了检测精度和速度，为割草机器人自动化提供了新途径。”摘要：There are some problems in lawn environment pedestrian detection model， such as low recognition rate， large model size， multiple parameters， and slow recognition speed， which make it difficult to deploy to robot platforms with limited computing power. A more lightweight with high-precision YOLO-CGO model depending on YOLOv5s is proposed to solve the above problems. First， the feature extraction network of the model was reset using the lightweight network MobileNetv3， reducing the number of model parameters and improving detection speed. Then improve the C3 module of the neck network by combining CA （Coordinate Attention） attention module. In the end replacing convolutional layers of the neck network with GSConv convolutional layers， and the last convolution layer was replaced by the ODConv convolution layer reduces the complexity of the model while maintaining accuracy. The experimental results show that the YOLO-CGO model designed in this paper on the self-built dataset reduces the parameter count by 38%， model volume by 38%， and computational load GFLOPS by 50% compared to the original model， achieving significant lightweighting； And compared with the original model， the model proposed in this article is superior in map@0.5 Up by 1 percentage point map@0.5 Increase by 1.7 percentage points above 0.95. This study indicates that the YOLO-CGO model proposed in the article can achieve excellent accuracy in extremely lightweight situations， providing a practical and feasible approach for the automation and intelligence of lawn mowing robots with limited computing resources.关键词：pedestrian detection;lightweight;GSConv;ODConv;CA attention module0|0|0发布时间：2024-12-30
- “在目标跟踪领域，研究人员提出了多滤波器支持相关滤波跟踪方法，有效利用样本历史信息，提高了跟踪精度。”摘要：In the support correlation filter tracking method， calculations are converted into frequency domain by cyclic sampling， which eliminates less sampling and high computational complexity of support vector machine （SVM）. However， the current method linearly interpolates historical and current samples during the tracking process to obtain training samples， which cannot effectively utilize the historical information of the samples. In response to this issue， this paper proposes a multi filter supported correlation filtering tracking method. During the tracking process， first train the historical filter using historical samples， and then use the historical filter to constrain the current filter， which can better utilize the historical information of the samples. Experiments on the OTB100 database showed that the algorithm achieved an accuracy of 79.2% and a success rate of 58.6%. Compared to the Scale Kernel Support Correlation Filtering algorithm （SKSCF）， the algorithm proposed in this paper improved accuracy and success rate respectively by 2% and 3.7%.关键词：multi-filter;SVM;support correlation filter;visual tracking4|0|0发布时间：2024-12-30
- “最新研究突破了脑神经元形态分析的瓶颈，提出了自适应网格划分方法，为脑区分层研究提供数据依据。”摘要：With the development of brain microscopy imaging technology， it is now possible to obtain a large amount of fine data on brain neurons. How to efficiently and accurately analyze the relationship between irregular brain regions and brain neurons is currently one of the bottleneck problems. The existing analysis methods mainly divide cortical brain regions into rectangular grid blocks， and analyze the morphology of neurons through the segmented regions. However， this method cannot maintain the topological structure of irregular brain regions after partitioning， resulting in difficulties in analyzing neuronal morphology. An adaptive grid partitioning method is proposed to address this issue. Based on the two-dimensional sequence contour map of brain regions， the grid shape is adaptively adjusted to match the geometric features of irregular brain regions， which can effectively surround the partitioned area. The experimental results show that this method can accurately divide irregular brain regions， maintain the topological structure of brain regions， and perform correct statistical analysis on neurons within brain regions， providing data basis for brain region stratification.关键词：adaptive grids;irregular brain regions;neurons4|0|0发布时间：2024-12-30
- “在视频超分辨率领域，研究者提出了一种基于可变形注意力空间变换的对齐方式，有效提升了低分辨率视频转高分辨率视频的性能。”摘要：Video super-resolution technology aims to convert low resolution videos into high-resolution videos. The existing feature alignment methods based on deformable convolution are limited by the receptive field size， and can only perform local offsets in the convolution space at specified spatial positions. The effect is not good when there is large-scale motion between frames. Therefore， a alignment method based on deformable attention space transformation is proposed to sample the entire feature map. Firstly， by offsetting， the sampling points are focused on any position related to the current processing location； Secondly， the model uses recursive structures to propagate fused features globally， and Transformer to extract features and align frames locally； Again， input the aligned features into a spatiotemporal feature fusion module with channel attention to supplement the reconstruction information； Finally， the output of the fusion module is propagated bidirectionally with a recursive network to supplement the temporal features of adjacent frames， and high-resolution video is obtained through sub-pixel convolution with 4x upsampling. The experiment shows that the network improves the PSNR index by 0.69 dB and 0.43 dB on the REDS4 and Vid4 datasets， respectively， with BasicVSR as the baseline.关键词：recurrent neural network;video super-resolution;Transformer;attention mechanism;deep learning4|0|0发布时间：2024-12-30
- “在计算机视觉领域，城市三维重建技术取得新进展。专家采用基于神经辐射场的阶段式三维重建方法，并提出基于远边界的采样分布策略，有效提升了渲染质量。”摘要：Urban 3D reconstruction is a hot issue that has received much attention in the field of computer vision. The wide application of urban 3D modeling covers many fields， however， there are still some problems in the current work of urban scene reconstruction. The problem of data multi-scale in urban scenes， the reconstruction of the near scene is blurred while the reconstruction of the far scene appears jagged， and the details of the far scene are under-represented and the edges are blurred. In order to solve these problems， a staged 3D reconstruction method based on neural radiation field is used and a sampling distribution strategy based on far boundary is proposed. The staged reconstruction method allows the model to learn the city scene layer by layer from far and near， which solves the multi-scale problem of city modeling. The sampling strategy is able to capture the details of the scene area more effectively by calculating the light distribution and sampling densely on the far boundary， which helps the model to learn and express the nuances of the urban scene more comprehensively and restore the details in the far boundary more accurately. Comparison in the experiment reveals that the image PSNR increased by 7.22%， SSIM increased by 17.20%， and LPIPS decreased by 32.40%， indicating that the method can effectively improve the rendering quality.关键词：neural radiation field;3D reconstruction;urban scene;sampling strategy;multi-scale data4|0|0发布时间：2024-12-30
- “在目标检测领域，FPCA-YOLOV5算法通过特征融合技术，有效提升了模型的检测精度和召回率。”摘要：A feature fusion algorithm FPCA-YOLOV5 with improved YOLOV5 is proposed to address the issues of missed detection of dense pedestrians and low detection accuracy. Firstly， by combining the spatial pooling pyramid structure SPPFCSPC with CA attention， the model has stronger expressive power. Secondly， adding PP modules to the network and changing the detection layer from three to four layers can achieve more accurate detection of small targets. Finally， a novel downsampling mechanism， CAConv， was designed to enable the network to focus more on important channels when processing feature maps. The experimental results show that on the public dataset WiderPerson， the improved YOLOV5 model has increased recall by 3.4% and average accuracy by 2.3% compared to the original model. The overall performance is significantly improved compared to the original model， demonstrating the effectiveness of the FPCA-YOLOV5 algorithm in object detection.关键词：YOLOv5;pedestrian detection;feature fusion;attention mechanism;small object detection4|0|0发布时间：2024-12-30
- “在目标跟踪领域，研究者提出了一种基于孪生网络的轻量级跟踪算法，优化了核心网络并引入注意力机制，有效提升了算法性能和鲁棒性。”摘要：Siamese network tracking algorithms can transform tracking problems into similarity matching problems， but most algorithms cannot be implemented in engineering applications on mobile devices or embedded devices with insufficient computing power. To this end， a lightweight tracking algorithm based on siamese networks is proposed， selecting ShuffleNetV2 as the core network that can be used on mobile devices. Aiming at the shortcomings of the original network， four optimization operations are proposed： eliminating the influence of padding layer， modifying activation function， adopting upsampling， and modifying step size. At the same time， attention mechanism is introduced to further strengthen the connection between features. Simulation experiments were conducted on the OTB100 and UAV123 datasets， and the results showed that compared with existing tracking algorithms， the proposed algorithm has excellent comprehensive performance. At the same time， it has good robustness in the face of various complex factors such as deformation， low resolution， and scale transformation.关键词：target tracking;ShuffleNetV2;Siamese network;attention mechanism4|0|0发布时间：2024-12-30
- “在体积渲染领域，实时体积路径追踪技术取得新进展，提出了自适应时空降噪方法，有效提升了图像质量与实时交互性能。”摘要：Volume Path Tracking （VPT） technology has been widely used in the field of volume rendering due to its ability to provide accurate global illumination effects. However， when generating rendered images， problems such as frame jitter， flicker， and noise may occur due to insufficient sampling rate or uneven sampling methods. Therefore， an adaptive spatiotemporal denoising method for real-time volume path tracking is proposed. Firstly， reuse available historical samples through time filtering， and use improved feature depth to guide sample reset and reduce ghosting phenomenon； Secondly， to prevent a decline in image quality after sample reset， improved adaptive sampling is used to dynamically adjust the sample size， balancing real-time interactive performance while maintaining image quality； Finally， spatial filtering is used to remove noise in the image that cannot be processed by temporal filtering， in order to further improve the denoising effect. The experiment shows that compared with ADTS， SVGF and other methods， the proposed method improves PSNR and SSIM by 0.79 dB and 2.08%， and 4.12 dB and 3.98%， respectively， while preserving more image details in visual effects.关键词：volumetric path tracing;denoiser;adaptive sampling;three-dimensional reconstruction;real-time rendering4|0|0发布时间：2024-12-30
-
Research on UAV Image Semantic Segmentation Algorithm for Vegetation Protection in Offshore Areas AI导读
“在海洋生态保护领域，专家设计了结合CNN与Transformer的特征提取网络，构建了MEAFormer分支和类引导加权模块，为无人机海岸带植被分割提供了有效策略。”摘要：Coastal zones have a profound impact on human life and economic development. UAV has been widely used in marine ecological protection. However， existing segmentation models still have some problems in UAV coastal zone vegetation segmentation tasks. Therefore， this paper designs a feature extraction network combined with CNN and Transformer， and then designs a MEAFormer branch. Meanwhile， a class-guided weighting module （CGW） is designed to learn the robust feature representation of different appearances. On the other hand， there are similar vegetation category segmentation errors and unclear underwater vegetation boundary segmentation caused by coastal zone scenes. Therefore， this paper constructs a fusion branch including mixed Convolution attention module （MCA） and dual attention fusion module （DAFM） to integrate and learn features of different levels. Meanwhile， SAM branch is introduced. The Mask obtained by the MEAFormer branch guides SAM to do fine segmentation. The MIou achieved 79.8% on Cityscapes and 72.4% on OUC-UAV-SEG， respectively， the effectiveness of the segmentation strategy proposed in this paper was verified.关键词：vegetation protection;unmanned aerial vehicle image segmentation;class guided weighting;feature fusion4|0|0发布时间：2024-12-30
Graphic and Image Processing
- “在智能审计领域，南京审计大学计算机学院产教融合模式培养复合型人才成效显著，为高等教育机构提供借鉴。”摘要：In the contemporary era of rapid information technology development， intelligent auditing has gradually become an innovative trend in the field of auditing. This study aims to explore the effective models for cultivating innovative talents in intelligent auditing under the background of industry-education integration， taking the practice of the School of Computer Science and Technology at Nanjing Audit University as an example， and analyzes its effectiveness and experience in training compound talents needed in the new era. The results show that the model has significant effects in promoting students' innovative thinking and practical abilities， and it can realize the deep integration of disciplines and the comprehensive integration of curriculum resources， the strengthening of the construction of internship and training platforms and the comprehensive cultivation of innovation ability， the careful construction of dual-teacher teams and the continuous improvement of teaching quality， and the remarkable improvement of students' innovation ability and employment competitiveness， it has reference significance for other higher education institutions.关键词：industry-education integration;intelligent auditing;talent cultivation;information technology4|0|0发布时间：2024-12-30
- “在教育领域，虚拟现实技术、ChatGPT、Sora等新兴技术推动了融合式教学新形态的产生，提出了培养学生高阶思维和迁移能力的新要求。研究构建了面向深度学习的融合式教学实施框架，验证了其可有效提升学生核心能力与体验成效，达成高质量教学与学习目的。”摘要：Emerging technologies such as virtual reality， ChatGPT， and Sora have brought strong impetus and support to teaching， promoting the emergence of new forms of integrated teaching. At the same time， they have also put forward new requirements for talent cultivation， namely cultivating students' core abilities such as higher-order thinking and transfer. On the basis of literature review， a deep learning oriented integrated teaching implementation framework was constructed， consisting of a deep learning objective module （ability layer， experience layer， and effectiveness layer） and an integrated teaching implementation module （technology layer， learning activity layer， and guidance activity layer）. The framework was validated from five dimensions： core literacy ability， metacognitive ability， learning behavior， teaching activities， and satisfaction， and was found to effectively promote the improvement of students' core abilities and experience effectiveness， achieving the goal of high-quality teaching and learning. The integrated teaching framework for deep learning provides a reference for the implementation of intelligent supported deep learning and integrated teaching.关键词：deep learning;integrated teaching;intelligent technology;core competence;experience effect4|0|0发布时间：2024-12-30
- “量子信息科技发展迅速，高校开设量子信息课程至关重要。虚拟仿真实验教学提升学生成绩和实践能力，优秀学生比例显著增加。”摘要：In the context of the information age， the rapid development of quantum information technology has a profound impact on technological innovation. Therefore， it is crucial for universities to integrate virtual simulation experiments and offer courses on quantum information to cultivate students with innovative spirit and scientific exploration ability. Firstly， the importance of offering such courses is discussed， and the overall architecture， main modules， and scoring features of the course are introduced using quantum entanglement in the virtual simulation experiment teaching offered by our own team as an example； Then， conduct a thorough analysis of the advantages of utilizing virtual simulation experiments to enhance teaching effectiveness and students' practical abilities. Practice has shown that the proportion of excellent students has increased from 7.1% to 24.9% after adopting virtual simulation teaching， and the proportion of students with passing grades or below has decreased from 25.3% to 10.2%. This proves that the virtual simulation teaching method of quantum entanglement can not only effectively improve students' grades， but also highlight its innovative potential in the field of education.关键词：virtual simulation;quantum information;quantum entanglement4|0|0发布时间：2024-12-30
Computer Talents and Education
- Technical support is provided by Beijing Founder electronics co., LTD 鄂ICP备18021494号-4
鄂公网安备42010602004949号 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0