最新刊期
卷 21 ， 期 4 ， 2022
- 摘要：Machine learning has made great achievements in healthcare in the past few years， however， these methods are only based on correlation， not causation. The particularities of the healthcare determines that the research method must comply with the causality norm， otherwise the wrong intervention measures may bring the patients a lifetime of misfortune. Therefore， we propose a method of causal feature discovery and prediction for instance disease prediction. Causal feature selection is based on function causal generation neural network， and use reinforcement learning framework to design interpretable causal instance feature selection prediction.The model is composed of three neural networks （causal prediction network， baseline prediction network and causal gate selection network）， and the Actor-Critical method is used to train the network. The results of our experiments on synthetic data， open source data and real medical data show that our proposed method can provide causal explanations for the model while giving prediction results.关键词：causal effects;interpretability;feature selection;disease prediction
- 183
- |
- 137
- |
- 0
发布时间：2022-04-26 - 摘要：Medical data is a type of big data， which has the characteristics of large data volume， rapid growth， multi-modality， high value， and emphasis on privacy. According to the characteristics of medical data， how to organize and manage it uniformly and efficiently is an important research direction at present. Therefore， by expounding the characteristics of medical data， analyzing its storage requirements， combing the application status of distributed system and non relational database in the field of medical data storage， this paper comprehensively analyzes the key core problems to be solved in the future， in order to explore the medical data storage scheme with high efficiency and fast response.关键词：medical data;storage solution;distributed system;NoSQL database
- 99
- |
- 16
- |
- 0
发布时间：2022-04-26 - 摘要：The automatic segmentation and extraction of glomeruli in kidney pathological images is very important for the rapid diagnosis of kidney disease. Aiming at the problem of insufficient high-quality labeled glomerular pathological image data， a glomerular pathological image synthesis method based on pix2pixHD is proposed. First， extract the glomerulus and its corresponding mask in the original pathological image to construct a glomerular pathological image data set； secondly， use the glomerular mask as a constraint condition to synthesize a high-quality kidney using the constructed generative adversarial network model The glomerular pathological image； finally， the generated glomerular image data set is merged into the original data set， and the U-Net segmentation model is used to re-segment the original kidney pathological image. Experimental results show that comparing different generation algorithms， the method in this paper performs best， in which IS±std is 1.50±0.03 and FID is 33.13； in the case of the same test set， the overall segmentation accuracy of this method is improved to 4%.关键词：generative confrontation network;U-Net network;glomerulus;image segmentation;pathological image
- 89
- |
- 18
- |
- 0
发布时间：2022-04-26 - 摘要：In order to solve the problem that heartbeat signal is easy to be disturbed by respiratory harmonics and other noise， a clustering ensemble empirical mode decomposition （EEMD） sign extraction model combining genetic algorithm （GA） and back propagation （BP） neural network is proposed. Firstly， moving targets detection （MTD） is used to filter the static clutter from the echo signal received by ultra wide band （UWB） radar； Secondly， the body surface vibration signal is extracted by the distance gate selection method， and the intrinsic mode functions component is obtained by EEMD decomposition of the body surface vibration signal； Finally， the weight of the feature vector transformed by intrinsic mode functions component is trained by GA-BP neural network， and Bayesian regularization is used as the training function of BP to reconstruct the cardiopulmonary signal， which is compared with the original EEMD reconstructed signal. The simulation results show that under different signal-to-noise ratios， signals extracted by GA-BP neural network has higher consistency with the actual results， and can effectively improve the extraction accuracy of respiratory and heartbeat signals.关键词：ultra wide band radar;ensemble empirical mode decomposition;GA-BP neural network;Bayesian regularization;signal processing
- 32
- |
- 8
- |
- 0
发布时间：2022-04-26 - 摘要：Knee osteoarthritis has a high disability rate，it is very important to take appropriate strategies to effectively control it.A rotation invariant nonlinear local fuzzy coding is presented in this paper. According to the grid comparison of neighbor pixels and the seed pixel in local unit， the authors obtain the exponential membership function by fitting envelope curve of gray level differences， and the local coding characteristic is described with the membership of pixel difference. Then the local fuzzy coding is depicted in the frequency histogram of the number of fuzzy coding units. By the enlightenment of shifting， the coding unit vectors， same after rotation， are merged as one class， which makes the dimension of vector describing the local fuzzy coding decreased. After that， a local fuzzy coding descriptor is proposed to reduce the calculation further. Experimental results have shown that compared with local fuzzy coding， the maximum accuracy of clustering is improved by 9% and 1%， the minimum accuracy is improved by 6% and 3%， and the average accuracy is improved by 14% and 7%， while the maximum accuracy of clustering is improved by 13% and 5%， the minimum accuracy is improved by 22% and 17%， and the average accuracy is improved by 21% and 9% on the Brodatz dataset respectively. The method proposed in this paper provides much better rotation invariance and anti-noise property， which can be extended to aided medical diagnoses of knee osteoarthritis.关键词：texture feature;local fuzzy coding;knee osteoarthritis;aided diagnoses;wise information technology of medicine
- 22
- |
- 7
- |
- 0
发布时间：2022-04-26 - 摘要：Medical staff often rely on personal experience to give solutions when dealing with medical adverse events. The algorithm adopts the combination of knowledge graph feature learning and collaborative filtering algorithm to collect and train the long-term accumulated adverse event data sets in hospitals， so as to provide the optimal solution for medical staff. Firstly， the information of patients in adverse events was transformed into low latitude vector by the knowledge graph training algorithm， and the similarity matrix of patients was calculated by the cosine similarity algorithm. Then the similarity matrix of adverse events is calculated by collaborative filtering algorithm depending on the basic information of the events. Finally， the two are combined into a recommended result set. The accuracy of the recommended result set provided by this algorithm is 35% higher on average than that of the traditional method， which not only reduces the burden of the medical staff， but also enables the management to trace the responsible person more effectively.关键词：knowledge graph;collaborative filtering;semantic similarity;recommender system
- 26
- |
- 11
- |
- 0
发布时间：2022-04-26 - 摘要：Due to the over segmentation phenomenon of watershed algorithm in human spine image segmentation and the problem of being very sensitive to weak edges and noise， this paper improves it. There are many unnecessary local minima in the original human spine CT images. Then， morphological image processing technology is used to denoise the initial segmented image to make the image smooth； The maximum value of the extracted region is marked as the foreground of the image， and the image obtained by threshold segmentation is marked as the background. Finally， the segmentation result of human spine is obtained by watershed transform. Experiments show that the algorithm can achieve accurate segmentation of human spine image. The mean values of Dice coefficient， Jaccard coefficient and precision coefficient are 89.2%， 82.3% and 85.4% respectively， which are 15.2%， 12.7% and 10.3% higher than the current mainstream algorithms.关键词：watershed algorithm;K-means++;human spine image;image segmentation
- 19
- |
- 7
- |
- 0
发布时间：2022-04-26 - 摘要：In the entire medical image processing process， image feature extraction is a very important step， which provides a strong foundation for disease diagnosis. In this paper， a medical image processing platform is designed， which can realize the basic operation of image data， extract the characteristics of cancer lesions， and provide data support for subsequent work. The platform adopts the MVT development model， the front-end design mainly adopts HTML5 technology， and the algorithm-level design introduces Python-related medical image processing algorithm packages to provide algorithm support for image processing. The B/S architecture is used to realize the linkage between front-end display and back-end data， forming a complete medical image processing system. In order to verify the availability of the features extracted by the platform，we take breast cancer as an example， combined with a machine learning related binary classification model for verification. Through the integration of algorithms with different functions， the developed platform can run on the computer normally， also the basic operation of medical images and the extraction of image features can be performed on the platform，which can provide data support for subsequent clinic work.关键词：medical image processing;platform design;Python;B/S architecture;characteristics calculation
- 12
- |
- 7
- |
- 0
发布时间：2022-04-26 - 摘要：To assist in the clinical diagnosis of gallbladder cancer，establish an auxiliary diagnosis model of gallbladder cancer based on patient enhanced CT images through improved 3D-DenseNet using deep learning technology， and discuss its clinical research value. Firstly， multiple arterial CT images of the patient are transformed into three-dimensional images， and the three-dimensional images are cut out of the region of interest by using the gallbladder region marked by the doctor. Then optimize the traditional DenseNet network， improve the Dropout mechanism and Softmax loss function， and replace the cross-entropy function with Focal-loss in the output part to perform imbalance correction， so as to establish the auxiliary diagnosis model of gallbladder cancer. Finally， by comparing the test set results with the gold standard， the ROC curve， recall and accuracy are calculated to evaluate the performance of the model.Through the continuous iterative training of the training set， the value of the model loss function converged and the diagnostic error decreased. After trying a variety of different model structures， the best model was selected， and the accuracy of the optimal model was 91.4%， the specificity was 95.2%， the sensitivity was 88%， and the accuracy was 95.8%.The gallbladder cancer diagnosis model based on 3D-DenseNet uses multiple CT image data of patients to extract depth features， and makes full use of the detection data of patients. The diagnostic model has high performance and accuracy， and can assist the clinical diagnosis of gallbladder cancer.关键词：gallbladder carcinoma;3D convolutional neural network;artificial intelligence;deep learning;diagnostic model
- 20
- |
- 9
- |
- 0
发布时间：2022-04-26 - 摘要：A daily attendance system has been issued by colleges and universities in view of the normalization trend of COVID-19 at present. While if foreign university staff needs to enter other universities for academic exchanges， this system is inconvenient for them. In the meantime， the incidents of health code fraud， epidemic rumors and so on have emerged over the years. In this thesis， it designs the college and university epidemic prevention and control system based on blockchain to solve the above problems. Besides， the corresponding Web interface is designed， which uses Hash algorithm to encrypt the plaintext of users. Meanwhile， it stores the epidemic-related data into the blockchain to solve the above problems. According to the system implementation and test results， it indicates that the designed epidemic management system by this thesis is able to effectively protect the user's personal travel privacy. On the other hand， it is able to trace the epidemic-related data， so that the purpose of avoiding the tampering can be realized. When it comes to the management system based on blockchain， it is able to both effectively protect user privacy and realize the traceability of epidemic data.关键词：epidemic management;blockchain;privacy protection
- 38
- |
- 15
- |
- 0
发布时间：2022-04-26 - 摘要：To improve the accuracy of early diabetes risk prediction，establish a diabetes risk prediction model based on an integrated learning algorithm.Using the early diabetes data set of the UCI database， a diabetes prediction model was established based on the ensemble learning method Random Forest， GBDT and XGBoost， and the prediction performance of 3 kinds of ensemble learning method and single classifiers such as SVM and BPNN classification were compared. Use the five indicators of accuracy， precision， recall， F1 score and AUC to evaluate the effect of the model， and rank the importance of variables based on the best performance model.Among all models， the XGBoost algorithm based on ensemble learning has the highest accuracy， precision， recall， F1 value， and AUC value， reaching 97.44%， 99.02%， 97.12%， 98.06%， and 99.41% respectively. Compared with SVM， the accuracy of prediction is increased by 3.85%， and the accuracy of prediction by BPNN is increased by 2.57%. The top ten factors influencing the weight ranking include polydipsia， polyuria， alopecia， weakness and sudden weight loss.The risk prediction models for diabetic patients based on these three integrated algorithms show better performance compared with single classifiers， can better identify early high-risk patients， and help clinicians make better medical decisions。关键词：diabetes;risk prediction;ensemble learning;SVM;BP neural network
- 14
- |
- 16
- |
- 0
发布时间：2022-04-26 - 摘要：Aiming at the deficiency of the traditional exponential smoothing time series model in epidemic prediction accuracy， and the traditional BP neural network is easy to fall into local optimization in the training process. Therefore， combining the particle swarm optimization algorithm （PSO） and BP neural network. A time series prediction of cumulative COVID-19 deaths and cumulative COVID-19 diagnoses in Minnesota from July 11 to August 25， 2021 was performed and compared with the traditional double exponential smoothing model （Holt） and BP neural network model. The experimental results show that the prediction accuracy of PSO-BP model is higher than that of the benchmark model， and the average percentage error （MAPE） is less than 0.08%. In six groups of prediction experiments， compared with Holt and BP models， the average error of PSO-BP model will be 0.038% and 0.012% lower， and the average absolute error （MAE） and root mean square error （RMSE） will be reduced by more than 5%.关键词：particle swarm optimization;exponential smoothing model;BP neural network;Covid-19
- 25
- |
- 16
- |
- 0
发布时间：2022-04-26 - 摘要：Understanding how multi-agent systems with many components reach a desired target is an important topic. Multi-agent system can reach the goals more effectively when the components are less capability. In order to give full play to its robustness， inspired by the olfactory and visual navigation modes of organisms， this paper uses group decision-making method to build a multi-agent navigation model. The complexity caused by multi agents’ collective feedback on information and actions drives the model organisms toward to the goal. The grid map is used to achieve the navigation algorithm based on biological olfactory heuristic， and the best parameters to make the entity as close as possible to the target are investigated in the environment with maze. And the experimental results show that achieving a balance between the sense of smell and vision is a critical point. Beyond this critical point， the possibility of an entity looking for food in a complex environment will greatly increase，it can cross obstacles to reach visible targets when the ability of agent is insufficient.关键词：multi-agent system;decentralized system;group decision-making;biologically inspired model;swarm intelligence
- 12
- |
- 7
- |
- 0
发布时间：2022-04-26 - 摘要：In order to determine early that the occurred earthquake is a strong earthquake， so that rescue operations can be carried out quickly and human lives and economic losses can be reduced， a model based on triplet convolutional neural network （Triplet-CNN） is proposed to investigate the prediction of strong and weak earthquakes. Taking the Miyagi region of Japan as an example， the historical seismic events in the region between 2000-01-01 and 2008-12-31， including the specific magnitude and the corresponding KiK-net and K-net strong-motor time series data， are obtained from the earthquake catalog of the region， and the designed Triplet-CNN structure is used to train the magnitude prediction model by combining these strong-motor data. The model is trained with the designed Triplet-CNN structure and these strong seismometer data to achieve fast determination of strong and weak earthquakes. Through the optimization of hyperparameters， the model achieves 96.85% accuracy， 96.83% precision， 96.82% recall， and 96.82% F1 value. Meanwhile， the Triplet-CNN classification model is compared with CNN classification model， random forest classification model and support vector machine （SVM） classification model， and the results show that the CNN seismic classification model has higher accuracy， precision， recall and F1 value， and the proposed Triplet-CNN-based seismic classification model can effectively and reliably make strong and weak earthquake prediction， thus aiding emergency decision-making and providing a favorable complement to earthquake early warning work.关键词：convolutional neural network;prediction of strong and weak earthquakes;Miyagi earthquake;strong motion seismograpn data
- 16
- |
- 4
- |
- 0
发布时间：2022-04-26 - 摘要：Crop recommendation based on soil data is one of the important contents of agricultural big data application. Aiming at the problems of the existing soil data recommendation model ignoring the soil text domain information， Insufficient ability to express cross-features of soil data and the indiscriminate treatment of cross features， a crop recommendation model based on attention and deep cross network （Attention Deep Cross Network， ADCN） was proposed. First， the soil text domain data is converted into low-dimensional dense vectors through vectorized embedding； then， the dense vector cross features are extracted through the attention mechanism and their weights are trained； second， the numerical domain features， dense vectors and cross features with weights are connected And input the cross network layer and the deep network layer； finally， combine the output of the deep network layer and the cross network layer to get the recommended crop. The data on the China Soil Dataset shows that the F1 of this model reaches 93.4%， which is better than the existing mainstream crop recommendation model and proves the superiority of the crop recommendation model integrated with the attention mechanism.关键词：crop recommendation;recommendation model;cross feature;attention mechanism;deep&cross network
- 17
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：Aiming at the low scalability of traditional methods for monocular 3D dense reconstruction tasks， based on the Transformer architecture， a 3D motion reconstruction network that can be solved by deep learning methods is improved， and the 3D reconstruction accuracy of the network is improved. Modify the query， key， and value matrix content of the Transfomer architecture， and use the dot product and Moore-Penrose pseudo-inverse operation to replace the dot product attention in Transformer. The process of traditional 3D non-rigid body motion reconstruction is embedded in Transformer to solve it. The experimental results show that the error rate of the 3D non-rigid body motion reconstruction network improved based on the Transformer architecture is close to the latest traditional 3D non-rigid body reconstruction method. Compared with the existing only 3D non-rigid body motion reconstruction network N-NRSfM network， the error rate is reduced by about 5% in some cases The dense 3D motion reconstruction network improved based on Transformer is the use of neural network to process monocular 3D non-rigid motion. Another attempt at rigid body motion reconstruction. Compared with the current only N-NRSfM network， the dense 3D motion reconstruction network based on Transformer is more reasonable in design and has a lower error rate.关键词：NRSfM;dense NRSfM;Transformer;computer vision;neural network
- 23
- |
- 16
- |
- 0
发布时间：2022-04-26 - 摘要：For the growing demand of different kinds of elements extraction in alarm text， a multiple model fusion method is proposed. For the irregular elements such as name， place and organization， the method of BERT + BiLSTM + CRF is adopted， This method extracts the key elements including semantic information by combining the context information of the text； for time， license plate number and other data with relevant rules， using the pattern recognition method， this method can extract the relevant elements in line with the defined rules， with a high recall rate； finally， the above two models are fused to form an integrated model for feature extraction. The experimental results show that compared with the others named entity recognition methods， the F1 value of BERT + BiLSTM + CRF model is improved by more than 3% and the pattern recognition effect is improved by more than 1%，which can meet the element extraction needs of daily police information.关键词：alarm text element;BERT;BiLSTM;CRF;named entity recognition;pattern recognition
- 13
- |
- 4
- |
- 0
发布时间：2022-04-26 - 摘要：Sales forecast is very important in the company's production and management， which affects the company's planning， production， logistics， sales and other aspects. To solve the limitation of single model for sales forecasting，this research proposes an Stacking model based on feature selection. It constructs new features based on historical features， and then uses correlation analysis to make feature selection， which improves the performance of the model.This paper uses random forest ，BP neural network ，Prophet algorithm as basic model， and linear regression as the final prediction model.The method is applied to the data set of kaggle’s sales forecasting competition， the results show that the MAPE of Stacking model is only 2.16%. Compared with the single model， the Stacking model reduces error by 1.42% at highest and 0.48% at lowest. Therefore， this method can effectively integrate the advantages of several models， make more accurate sales forecast， and provide strong support for enterprise decision-making.关键词：sales forecasting;time series;Stacking algorithm;feature selection
- 14
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：Industrial systems represented by train dispatch system of urban rail transit are often limited by project costs， and only two service nodes are deployed at the operation site. When center is offline due to a failure， if the mainstream distributed consistency protocols are used to ensure data consistency between nodes of operation site， availability will be lost and node failures cannot be tolerated. Therefore， this paper proposes a scheduling algorithm based on the Raft consensus protocol. Through reasonable expansion of the Leader election and data synchronization process， the router becomes the logical third node in the distributed system， meeting the majority design of the Raft protocol. Experiments have verified that the two-node distributed system based on this algorithm has improved availability compared to the two-node distributed system based on the Raft protocol and can tolerate the failure of a node. Compared with the three-node distributed system based on the Raft protocol， it is consistent in the correctness of leader election and data synchronization； limited by the number of nodes， the availability gap is 27.8%， and the leader election performance gap， which accounts for a small amount of running time， is 65.6%， The data synchronization performance， which accounts for the main running time， is increased by 7.8%. It can meet the project needs of a two-node distributed active-standby system and takes into account both availability and data consistency without increasing project costs.关键词：project cost;availability;data consistency;distributed system
- 18
- |
- 8
- |
- 0
发布时间：2022-04-26 - 摘要：Balanced Iteration Protocol and Clustering （BIRCH） algorithm using hierarchical method is a kind of hierarchical clustering algorithm， and also a kind of machine learning algorithm. With the development of machine learning， how to protect privacy in machine learning has become a hot research issue. Differential privacy is a new technology in data analysis and protection. One of its characteristics is that data can be compared after being processed by different parameters. Aiming at the privacy leakage phenomenon in machine learning， a differential privacy-oriented BIRCH（DPBIRCH） algorithm was proposed. Laplacian noise was added to the clustering feature tree to solve the privacy leakage phenomenon in the clustering feature. By analyzing the values of privacy budget parameters and the results of clustering accuracy in the experiments， DPBIRCH algorithm can provide different levels of privacy protection for data privacy within the range of privacy budget parameters and on the premise of ensuring clustering accuracy.关键词：machine learning;BIRCH algorithm;differential privacy protection;laplacian mechanism
- 21
- |
- 7
- |
- 0
发布时间：2022-04-26 - 摘要：Aiming at the problem that the load change of quadrotor aircraft affects the accuracy of trajectory tracking control， an adaptive trajectory tracking control strategy for the quadrotor UAV is proposed. A mathematical model of the quad-rotor aircraft is established according to the dynamics principle. On this basis， combined with the characteristics of under-actuation and strong coupling of the quad-rotor aircraft system， the control system is divided into an outer-loop position control subsystem and an inner-loop attitude control subsystem. Considering the load change of the aircraft when performing flight tasks， the aircraft mass is estimated according to the system state and compensated to the outer loop position control subsystem， combined with sliding mode control， an adaptive sliding mode controller is designed， and the Lyapunov theory is used to prove the stability of the control system Sex. Simulation experiments show that the proposed trajectory tracking control strategy has better tracking performance and robustness against interference.关键词：quadrotor UAV;trajectory tracking;self-adaptive;sliding mode control;anti-disturbance
- 13
- |
- 3
- |
- 0
发布时间：2022-04-26 - 摘要：Industrial Internet of Things （IIoT） is one of the application scenarios of wireless communication in the future，therefore investigate the feasibility of using cell-free massive multi-in multi-out （MIMO） with massive connectivity in IIoT. Specifically， in the uplink， cell-free massive MIMO with massive connectivity is used to support a large number of low-power IIoT devices. According to different cooperation degrees of Access Points （AP）， including centralized and distributed implementation， the spectrum efficiency and SNR of the system are analyzed by using different data processing methods. The simulation results showed that the centralized implementation using maximal-ratio （MR） processing has good performance and requires fewer forward signaling. The analysis scheme provided has some guiding significance for using cell-free massive MIMO with massive connectivity to serve IIoT.关键词：industrial Internet of Things;cell-free massive MIMO with massive connectivity;AP cooperation;MMSE processing;MR processing
- 9
- |
- 11
- |
- 0
发布时间：2022-04-26 - 摘要：In order to improve the efficiency of cloud computing task scheduling， reduce the cost and time of task scheduling， and overcome the shortcomings of algorithms that are easy to fall into local optimality， the w_BAPSO algorithm based on chaotic disturbance is proposed. In the cloud computing environment， linearly decreasing inertia weight with logistics chaotic disturbance is introduced. In the PSO update process， combining parameters such as bat pulse rate and pulse loudness， the w_BAPSO algorithm with chaotic disturbance is used in the cloud computing task scheduling process. The algorithm is applied to Cloudsim cloud computing simulation software for experimental verification. Test the effectiveness of w_BAPSO algorithm under different task scale conditions. Experimental results show that the chaotic disturbance w_BAPSO algorithm effectively reduces the time and cost of task scheduling in the process of cloud computing task scheduling. Therefore， the chaotic disturbance-based w_BAPSO algorithm proposed can not only improve the convergence speed and search accuracy of the algorithm， but also reduce the cost and time of cloud computing task scheduling， and improve the efficiency of task scheduling.关键词：cloud computing;task scheduling;particle swarm optimization;bat algorithm;chaotic disturbance
- 18
- |
- 6
- |
- 0
发布时间：2022-04-26 - 摘要：Aiming at the problems of incomplete acquisition of pulse signal， large volume and long time-consuming of pulse taking in single channel sampling of pulse diagnostic instrument， a three part pulse detection bracelet based on STM32 is designed to realize the three part measurement of pulse， and send data through bluetooth to meet the requirements of portability. The hardware design takes STM32f100 as the main control core， and the peripheral circuit is designed to collect the pulse change data. The upper computer uses LabVIEW and android as the platform to build the display program of PC and mobile terminal. By detecting the pulse of the same patient in the same period for many times， and taking the relevant indicators as a reference， the error before and after the equipment is finally measured to be about 2%. In addition， taking the standard pulse meter as the detection standard， the 95% confidence interval of the pulse rate of the system is ［- 2.783，2.803］. The pulse rate measurement results of the system and the pulse meter are highly consistent， indicating that the system has high reliability and good stability， and can be applied to pulse monitoring and research.关键词：STM32;three-position pulse diagnosis;piezoresistive sensor;bluetooth
- 21
- |
- 6
- |
- 0
发布时间：2022-04-26 - 摘要：In order to reduce the disability rate and mortality rate of the elderly caused by falls， collect motion data through wearable devices， combine the advantages of threshold method and pattern recognition algorithm， propose a simultaneous discriminant fall detection algorithm based on posture fusion， and introduce cloud technology and WeChat small program to design a convenient and feasible fall detection system. Firstly， the original data is preprocessed by complementary filtering to solve the problem of inaccurate attitude calculation caused by poor low-frequency characteristics of gyroscope and poor high-frequency characteristics of accelerometer. Then， the suspicious fall behavior is judged by threshold method to filter out most of daily activities. When suspicious behavior is detected， SVM algorithm is started for fall recognition. The experimental results show that the accuracy， sensitivity and specificity of the algorithm are 90.1%， 91.2% and 89.4% respectively， and the missing and false alarm rates are only 8.8% and 10.6% respectively.关键词：fall detection;attitude fusion;complementary filtering;gyroscope;accelerometer;SVM
- 14
- |
- 6
- |
- 0
发布时间：2022-04-26 - 摘要：Teleoperation interventional surgery is an important method for the treatment of cardiovascular diseases. How to analyze and evaluate the transparency of teleoperation system is an important problem. Aiming at the problem that the transparency evaluation method of teleoperation system is not enough and easy to understand， a dynamic model of the master-slave robot system is established， and a quantitative evaluation standard for the transparency of the system is made from three aspects： working frequency， virtual environment force perception threshold and system response time. Then， the sliding mode control algorithm of the delayed output observer is designed based on the bilateral teleoperation system model. The simulation results show that the sliding mode control algorithm of the delayed output observer can reduce the time delay of the system and improve the effect of force control. Meanwhile， the rationality of the proposed transparency evaluation standard is verified.关键词：teleoperation system;transparency;interventional surgery;time delay observer;sliding mode control
- 9
- |
- 3
- |
- 0
发布时间：2022-04-26 - 摘要：Teaching quality is the lifeline of colleges and universities. Therefore teaching evaluation is an important means to strengthen teaching management and constantly improve teaching quality. With the rapid change of information technology and the normalization of COVID-19 prevention and control， it has become an urgent need to develop an online teaching quality evaluation system that is portable and intelligent， practical and improves work efficiency. The system builds personal mobile device based on WeChat mini program and builds the content manage system based on "Cloud Base + Python Flask". The system realizes the online teaching evaluation system with complete data chain from curriculum schedule query to teaching evaluation， and then to feedback. The system has been put into use in Hainan Medical University， running stably， meeting the needs of multi-scene class evaluation， improving the efficiency of quality control management， and promoting the development of teaching quality.关键词：teaching evaluation system;WeChat mini program;Cloud Base;Flask
- 40
- |
- 13
- |
- 0
发布时间：2022-04-26 - 摘要：There are few studies on CAD software for circular knitting underwear machines nowadays， and most of the existing CAD softwares are focused on the PC side， which is costly to maintain and inefficient to run. Therefore， apply Internet technology to develop a online fabric CAD system specifically for circular knitting underwear machines， providing pattern design and chain-list action process design， and generating upload files to guide underwear machine knitting production. The system satisfies the user to access the system anytime and anywhere， realizing zero installation and maintenance of the client. Fabric design examples have proven that fabric design using this system is simple and efficient. The system has high practical application value in the field of circular knitting lingerie machines and is of great significance to the development of knitting CAD technology.关键词：circular knitted underwear machine;CAD;Web side;pattern design;chain-list action design
- 14
- |
- 3
- |
- 0
发布时间：2022-04-26 - 摘要：Aiming at the problem that traditional cloud service recommendation algorithm only considered the recommendation precision but ignored the recommendation efficiency， the cloud service recommendation algorithm considering MeanShift user clustering is proposed. MeanShift clustering algorithm requires a little computation，runs fast and can estimate the density of arbitrary distributed data. The cloud service recommendation algorithm first accesses all users according to the data density distribution. Then it calculates the cumulative frequency of users in each class cluster and puts respectively them into the class cluster with the largest cumulative frequency. Finally， the nearest neighbor set of the target user in the class cluster can be used for recommending cloud services. The sample recommendation results show that the proposed algorithm reduces mean absolute error by about 10.15%， compared with the trust-cloud recommendation algorithm， trust-cloud hybrid recommendation algorithm， and the hybrid algorithm based on grey correlation and trust-cloud. The root-mean-square error was decreased by about 7.87%. The execution time was reduced by about 59.77%. It shows that the proposed algorithm can effectively improve the recommendation efficiency on the basis of ensuring certain recommendation accuracy.关键词：cloud service recommendation;MeanShift;grey correlation prediction;trust cloud
- 13
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：Affected by the new COVID-19 epidemic， the application of online online courses in teaching has become more and more extensive， and how to successfully carry out online courses in colleges and universities has become a current hot issue. Use Python to analyze the learning data based on the students' online course learning situation in a university in Jiangsu Province. Find out the factors that affect the performance of online courses， and lay the foundation for better development of online course teaching in universities. Referencing the analysis and conclusions will not only help schools adjust the professional direction of online courses， rationally allocate teaching resources and improve teaching quality， but also provide a useful reference for other college students to establish correct learning attitudes and appropriate online courses.关键词：college teaching;online courses;data analysis;Python
- 24
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：In order to provide recommended codes of the international standardized framework ICF for rehabilitation practitioners when they use rehabilitation scales， uses text similarity to achieve ICF mapping of rehabilitation scales.Using the Chinese text of ICF and 6 stroke rehabilitation scales as data， and uses WMD， an unsupervised text similarity algorithm between documents， to map the rehabilitation scale items to ICF categories. The stroke comprehensive in ICF core sets is taken as the evaluation standard， and top-1 and top-5 accuracy are evaluated respectively. At the same time， the word vectors are regularized by L2 norm. The results show that the method proposed has an average top-1 accuracy rate of 0.9， a top-5 accuracy rate of 1， and an average of 3 correct answers according to the top-5 codes in the stroke comprehensive. The results show that the method proposed can effectively realize the ICF mapping of the rehabilitation scale， and it provides a new practical idea for helping rehabilitation practitioners to apply ICF.关键词：text similarity;rehabilitation scales;ICF mapping;WMD
- 9
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：In order to improve the effect of credit evaluation classification， a personal credit evaluation model combining cost sensitivity and integrated algorithm is presented to solve the problem of continuous and discrete co-existence and class imbalance. The optimal feature subset is generated by the feature selection algorithm of integrated information value， mutual information， information gain rate and gini index， and the integrated model based on L1-Logistic Regression， Elastic Networks -Logistic Regression， Bayes， Decision Tree and Neural Network is constructed by combining cost sensitivity， so as to realize the dynamic weighted voting strategy. The empirical results show that， compared with the original feature set， the G-means and F-value of the model after the feature selection algorithm are increased by 8% and 18%. Compared with the prediction performance of the model after the mRMR feature selection， the model also has a certain improvement， and showed certain robustness.关键词：credit evaluation;price sensitive;integration algorithm;feature selection;machine learning
- 10
- |
- 4
- |
- 0
发布时间：2022-04-26 -
Application Research on Weighted Gene Co-expression Network Analysis in Gene Screening of Sarcopenia AI导读
摘要：In order to solve the limitations of few therapeutic targets and gene screening methods for sarcopenia， a weighted gene coexpression network was proposed to identify sarcopenia genes. The coexpression network is constructed based on the topological matrix， the similar genes are cut into gene clusters by hierarchical clustering method and dynamic shear tree method， the eigenvalues of each gene module are extracted by principal component analysis， the correlation between the eigenvalues of gene modules and the trait vector of sarcopenia is calculated， the modules highly associated with sarcopenia are screened， and their hub genes are selected for verification. The experiment shows that 86 hub genes can be screened from 16，879 genes， and 69 are successfully verified. The accuracy is 80.2%， which is about 40% higher than that of traditional methods.关键词：co-expression network analysis;data mining;sarcopenia- 9
- |
- 3
- |
- 0
发布时间：2022-04-26 - 摘要：In order to solve the problems that the execution process of collaborative business process based on blockchain is difficult to support large-scale data storage and lacks of dynamic traceable access control scheme， a collaborative business process management framework is designed. Based on the blockchain smart contract technology， the framework adopts the interstellar file system technology to support the storage of data files under the chain， and combines the task and attribute access control model to support the traceable， fine-grained and dynamic data access control. The test results of a real case from the hearing aid industry show that the Gas consumption of the framework deployment cost， data import cost， single instance creation and execution cost， and single instance file management and access control cost are 11 673 340， 10 091 757， 2 880 868 and 2 267 735，respectively， which can support the shared storage and access control of data files in the process of collaborative business process execution， and achieve the expected design goal.关键词：business process management;blockchain;smart contract;data sharing;access control
- 17
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：To improve the problems of information asymmetry and opaque activity process caused by the centralized management of traditional public welfare activity platforms， the blockchain technology is applied to the management of public welfare activities. Taking advantage of the natural advantages of blockchain technology， such as decentralization， transparency， tampering proof and traceability， a blockchain-based platform for public welfare activities is designed， and the operating mechanism of the platform is explained. It's implemented through smart contracts in the conflux test network， and the technical feasibility is verified. The blockchain-based public welfare activity platform can increase the transparency of activities and help promote people's enthusiasm to participate in public welfare activities.关键词：blockchain technology;smart contract;Conflux;public welfare activity;transparency
- 24
- |
- 4
- |
- 0
发布时间：2022-04-26 - 摘要：Writing personal information on the courier bill will not only lead to personal information leakage， but also prone to fraudulent delivery of couriers， and couriers who claim to sign for damaged couriers without authorization. Due to the lack of a complete tracing mechanism， it is impossible to determine the responsible party for the problem， and it is easy to shirk each other， which affects the customer experience. The courier company centrally stores the user's personal information in the database. When the database is attacked， the user's data will be exposed to the risk of leakage. Based on the above situation， design privacy protection system for personal express mail based on blockchain technology. The user encrypts the order information through the APP and uploads it to the database. The encrypted express order is displayed in the form of a QR code. Even if the information is leaked， others cannot identify it； All information is on the blockchain， so that the entire transportation process can be traced and cannot be tampered with； when the express is lost or damaged， you can apply for compensation immediately； the APP uses an elliptic curve encryption algorithm， and the private key cannot be derived from the public key， so only the user can sign for express delivery， thereby reducing the phenomenon of signing on behalf of the person， wrong receipt， etc.关键词：blockchain;QR code;privacy protection
- 21
- |
- 6
- |
- 0
发布时间：2022-04-26 - 摘要：Impervious surface is an important factor affecting the ecological environment of mountainous areas. Multi-source remote sensing data fusion is an important method for impervious surface extraction， but it is easy to cause classification features redundancy and feature optimization is needed. Existing researches lack comparative analysis of different feature optimization methods in remote sensing extraction of impervious surface. Firstly， we used Sentinel-1 and Sentinel-2 images extracted features. Then， features were optimized based on Jeffries-Matusita distance， Random Forest （RF） model and ReliefF algorithm respectively， and the classification accuracy of different feature subsets were evaluated based on RF classification model. Finally， the impervious surface of mountainous city Kunming in 2020 was extracted. The results show that texture features were more important than spectral and topographic features. The feature subset based on ReliefF algorithm had the highest classification efficiency and accuracy， and the training time of the model was reduced by 84 seconds and 16 seconds respectively compared with JM distance and RF， with an overall classification accuracy of 0.95 and a Kappa coefficient of 0.79. This article provides a reference for remote sensing extraction of ground objects based on multi-source data.关键词：feature optimization;Sentinel images;mountainous city;impervious surface;random forest
- 35
- |
- 5
- |
- 0
发布时间：2022-04-26 - 摘要：Aiming at the problem of slow target detection speed and difficulty in detecting small targets in pedestrian target detection tasks， a YOLOv3 multi-scale target detection model fused with CBAM attention mechanism is proposed. The algorithm first uses YOLOv3 as the basic network for feature extraction， and then adds a double down-sampled feature map to the multi-scale feature fusion layer of YOLOv3 to supplement the feature information of small targets. Finally， it is fused in each scale feature map of YOLOv3 Later， the Convolutional Block Attention Module （CBAM） is added to strengthen the feature expression ability of the network. The DIOU loss function is used to replace the more mainstream GIOU loss function during model training. The model uses the INRIA data set for experiments. The experimental results show that the improved YOLOv3 target detection model has a greater accuracy and speed， and the detection accuracy is maximized by 4.5%， the detection speed is increased by 8 frames/s， which verifies the feasibility and effectiveness of the model.关键词：pedestrian target detection;multi-scale feedback;CBAM attention mechanism;DIOU loss function pedestrian
- 43
- |
- 9
- |
- 0
发布时间：2022-04-26 - 摘要：In order to solve the problem of difficult and single expression feature extraction in expression recognition task， an expression recognition method based on the combination of deep residual network and geometric features is proposed. In this method， the deep residual network is improved by using continuous small convolution instead of large convolution， adding attention mechanism to each residual block， taking the same two improved residual networks as branch networks， extracting expression features respectively， and then aggregating them by adding and averaging to achieve feature complementarity and get global features. Face key points are used to construct and calculate face geometric features. Principal component analysis was used to remove redundant information to get geometric features， which are fused with global features as the final expression features. Softmax is used for classification. The experimental results on CK+ and JAFFE datasets show that the recognition rate of the improved method is 98.05% and 95.29% respectively. Compared with some common facial expression recognition methods， the accuracy of expression recognition method based on deep residual network and face key points are improved to a certain extent.关键词：facial expression recognition;residual network;attention mechanism;Geometric features
- 19
- |
- 10
- |
- 0
发布时间：2022-04-26 - 摘要：Human action recognition is a research hotspot in the field of artificial intelligence， compared with video， motion stream data， human skeleton data has simplicity and high efficiency of vector calculation. The related works of human action recognition based on skeleton data are reviewed from two aspects： manual feature extraction method based on traditional machine learning and deep feature extraction method based on deep learning. The manual features are summarized as physical attribute features and statistical attribute features， according to the depth feature， the human action recognition method based on skeleton data and its advantages and disadvantages are described one by one according to convolution neural network， recurrent neural network， graph convolution network and hybrid neural network. The feature extraction method， technical route， model characteristics and recognition rate are analyzed and summarized in tabular form.关键词：human action recognition;skeleton data;manual feature;deep feature
- 13
- |
- 7
- |
- 0
发布时间：2022-04-26 - 摘要：At present， China's digital home is evolving towards the smart home， in which accurate recognition of family members' action categories plays an important role. This paper firstly reviews the action recognition algorithms for human body in smart home， describes the mainstream algorithms in the field of action recognition based on traditional methods and deep learning （including Transformer）， points out the shortcomings， and summarizes the current main datasets in the field. Next， the application process of action recognition in specific scenarios in smart homes is indicated， and the value of human action recognition research in smart homes is analyzed. Through the above research， it is found that there are limitations of existing action recognition algorithms in the application of smart home construction，new motion recognition models and methods still need to be explored.关键词：smart home;action recognition;feature extraction;Transformer
- 22
- |
- 11
- |
- 0
发布时间：2022-04-26 -
Comparative Study of Denoising Methods Based on Head Motion Correction in Resting-State fMRI of Rats AI导读
摘要：In neuroscience research， resting-state functional magnetic resonance imaging （rs-fMRI） not only reflects neural activity， but also preserves the noise caused by head movement， physiology， and hardware， which leads to erroneous estimates of resting-state functional connectivity. In order to obtain an accurate functional connectivity network， this paper analyzes the source of noise， introduces motion correction methods， and studies denoising methods such as independent component analysis， frequency filtering， and CompCor signal regression and their applications in rodents. By evaluating the signal-to-noise ratio （SNR）， we found that images denoised by CompCor and independent component analysis have higher SNR and can be an alternative to controversial denoising methods. Although denoising has been developed relatively rapidly， noise is complex and interacts with each other， and eliminating noise to the greatest extent without affecting the signal of the region of interest is still the research direction of static state data preprocessing.关键词：functional magnetic resonance imaging;resting state;denoising method;rodents- 11
- |
- 6
- |
- 0
发布时间：2022-04-26
- Technical support is provided by Beijing Founder electronics co., LTD 鄂ICP备12016542号-70
京公网安备11010802024621 - 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