最新刊期
卷 21 ， 期 8 ， 2022
- 摘要：The college entrance examination reading comprehension test is a very difficult research object in the field of machine reading comprehension because of its high language complexity and great difficulty in automatic answer. Essay question is an important type of reading comprehension questions in the college entrance examination. The answers are usually distributed in different paragraphs of reading materials. The existing extraction reading comprehension methods do not consider this situation. At the same time， the number of answer sentences in the materials is usually much smaller than that of non answer sentences， which is unbalanced. Based on this， the answer frame of "paragraph filter - answer sentence extraction" is adopted to calculate the correlation score between paragraphs and questions， sort the paragraphs， select the most relevant first ［k］ paragraphs for each question， and use the data enhancement method to expand the answer sentences in the data set to solve the problem of fewer answer sentences. The experimental results show that the recall rate of the proposed method is 51.76%， which is 11.37% higher than the baseline model RoBERTa.关键词：answer sentence extraction;machine reading comprehension;essay question;natural language processing
- 151
- |
- 32
- |
- 0
发布时间：2022-08-18 - 摘要：As network intrusion behaviors become more diversified and intelligent， network data feature dimensions are high and non-linear and separable， traditional intrusion detection algorithms have problems such as insufficient extraction of network data features and low model classification accuracy. In response to the above problems， use a combination of Convolution Neural Network （CNN） and Bi-directional Long Short-term Memory （BiLSTM） to extract features from the network data， and through multiple feature extractions Construct a multi-granular feature space. And use the classifier based on the Three-way Decision （TWD） theory to classify the network behavior， and further analyze the network behavior divided into the boundary domain according to the characteristics of different granularities. Therefore， an intrusion detection method based on CNN-BiLSTM and three-branch decision is proposed. Experiments on NSL-KDD data set and CIC-IDS2017 data set show that this method has better intrusion detection effect.关键词：CNN-BiLSTM;three-way decisions;feature extraction;intrusion detection
- 111
- |
- 29
- |
- 0
发布时间：2022-08-18 - 摘要：In order to solve the complexity and subjectivity of manual labeling of interictal epileptic discharge （IEDs） in clinic， deep learning method based on convolutional neural network is used to automatically extract EEG features. Based on the improved EEGNet model， deep and separable convolution layer is used to construct the model， which encapsulates the concept of EEG feature extraction. The experiment was performed on a self-collected dataset of real epileptic patients with interictal epileptic discharge labeling information. The proposed method can achieve the detection accuracy at 70.5%. The experiment shows that the method based on deep learning compact convolutional neural network has good detection performance for IED detection in epilepsy patients.关键词：electroencephalogram;convolutional neural network;deep learning;epilepsy;IED detection
- 77
- |
- 25
- |
- 0
发布时间：2022-08-18 - 摘要：Stance detection for user comments on social platforms aims to classify the stance of users' reviews towards a specific topic. Existing stance detection studies only focus on the internal semantic features of reviews text， while ignoring the external knowledge associated with the text of the review. Retrieve external knowledge related to the key information of a comment by mapping it to a knowledge graph and introduce the external knowledge into the model for the stance detection task， which can improve the classification result by providing background knowledge that may be crucial to the stance detection task. In addition to considering the textual features of reviews when constructing the stance detection model， employs a gated graph neural network approach to fuse the structural information between reviews which can capture the interactions of related reviews. The experiment shows that the model achieves better stance detection results compared with existing stance detection models for solving this problem. By combining the textual features and structural information of reviews and introducing external knowledge， the performance of the stance detection model can be effectively improved.关键词：stance detection;gated graph neural network;knowledge graph;structural information
- 58
- |
- 9
- |
- 0
发布时间：2022-08-18 - 摘要：Two most important models in the traditional sentiment analysis model are convolutional neural network （CNN） model and recurrent neural network （RNN） model， but CNN can only extract the local information of the text， and RNN is easy to fall into the gradient explosion problem. In response to the above problems， a method that combines CNN with two-layer bidirectional gated recurrent unit （BIGRU）is proposed. This method combines the advantages of CNN's ability to extract local features and two-layer BIGRU's ability to extract contextual semantics and enhance feature information. In addition， there are often linguistic irregularities in the text of sentiment analysis， which affects the accuracy of text extraction and analysis. In response to this problem， it is proposed to introduce the attention mechanism to the original word vector calculation model to focus on the importance of the text. information. Experimental results show that the accuracy of the model is 1.21% higher than that without the attention mechanism， 1.58% higher than the CNN-BILSTM model， and 1.39% higher than the CNN-BIGRU model.关键词：convolutional neural network;two-layer bidirectional gated recurrent neural network;attention mechanism;word vector;sentiment analysis
- 41
- |
- 11
- |
- 0
发布时间：2022-08-18 - 摘要：Because there was a strong nonlinear relationship between air quality and meteorological factors and many scholars only studied this problem based on a single method， select the air quality data of Shanghai from 2016 to 2021， and then built an air quality prediction model using BP neural network， decision tree and SVM to predict the air quality grade of the next day. The results show that： ①The model based on SVM has the highest prediction accuracy， cart decision is the second， and BP neural network is the worst； ②Among the four kinds of SVM models based on different kernel functions and classification methods， the prediction accuracy of SVM based on linear kernel function and one-against-the rest method is the highest， which is 80%； ③When the air quality is good， the coincidence between the predicted value and the real value is high. Applying machine learning method to air quality prediction is reasonable and can provide suggestions for citizens' travel.关键词：machine learning;BP neural network;SVM;decision tree;air quality
- 47
- |
- 13
- |
- 0
发布时间：2022-08-18 - 摘要：At present， intelligent inspection robots have been widely used in many fields of industrial production， the development of special inspection robots for special extreme scenarios is not yet mature，and there are bottlenecks in traditional robot fault detection，this paper aims to design a generator fault detection system for polar scientific research. The designed system firstly adopts the control strategy of self-organization and remote control to cope with the problems of poor network conditions and high communication cost in polar regions； secondly， the designed system proposes a small-sample fault identification algorithm based on siamese network， aiming to solve the problems of small sample size and unbalanced positive and negative samples in fault detection. The superiority of the performance of the proposed system can be verified through extensive system simulations.关键词：automatic inspection robot;polar research;small sample fault identification;artificial intelligence
- 13
- |
- 5
- |
- 0
发布时间：2022-08-18 - 摘要：Solving the recommendation of cold start users is a challenging problem in cross-domain recommendation. At present， most of the proposed cross-domain models use scoring information to obtain the potential characteristics of cold start users in the target domain， but the scoring information is relatively sparse. In order to better obtain the preference characteristics of cold-start users， this paper proposes a new cross-domain recommendation model DRCDR， which fuses comment information for the source domain and target domain at the same time. This model uses the information from two source domains to provide data for users. While dealing with the scoring information， it takes into account the context information contained in the temporal comments， and then synchronously adds the comment information to the source domain and the target domain. DRCDR also uses a new fusion method to fuse two source domain information， and uses a multilayer perceptron to obtain a non-linear mapping function between the source domain and the target domain to achieve cross-domain recommendation. The experimental results on real data sets show that the RMSE and MAE of the cross-domain recommendation method proposed are improved by 2%~4% compared with the previous cross-domain methods， and it can effectively realize cross-domain recommendation.关键词：transfer learning;Bi-LSTM;deep learning;cross-domain recommendation
- 30
- |
- 10
- |
- 0
发布时间：2022-08-18 - 摘要：In order to study the speech recognition of low-resource minority languages more effectively， propose a deep neural network model combining SE-Block and Inception_Resnet_v2. Inception structure is introduced in this model to extract spectral features， and SE-Block is introduced to enhance the ability of network to select information channels of feature dimensions and optimize network performance. Experimental results show that， compared with other mainstream models， the proposed model not only improves the model accuracy and training convergence speed， but also enhances the overall learning ability of the network. The experiment achieves good learning results on Wa language datasets， which verifies the high efficiency of the model in speech recognition of minority ethnic groups with limited data samples.关键词：isolated-word recognition;language spectra;Inception-ResNet_v2;SE-Block
- 16
- |
- 4
- |
- 0
发布时间：2022-08-18 - 摘要：Traditional book recommendation algorithms only consider external scoring data， and will face problems such as cold start of items. This paper combines book external scoring data with book connotation knowledge， and proposes a book recommendation algorithm （CKCF） integrating knowledge graph and collaborative filtering. Firstly， based on the training of knowledge graph， the semantic information of books is transformed into low-dimensional vector matrix， the semantic similarity between books is calculated by cosine similarity formula， the similarity between new books and other books is increased， and the semantic nearest neighbor of books is obtained according to the book semantic similarity matrix. At the same time， improve the collaborative filtering similarity calculation method， obtain the book scoring nearest neighbor according to the book external scoring matrix， and finally combine the scoring nearest neighbor and semantic nearest neighbor to obtain the final book recommendation result set. The algorithm is tested on the Book-Crossing dataset. The experimental results show that the accuracy of the algorithm is improved to 4.37%， 0.69% higher than the traditional method， and has better performance than other related algorithms.关键词：collaborative filtering;knowledge graph;semantic similarity;recommendation system
- 44
- |
- 11
- |
- 0
发布时间：2022-08-18 - 摘要：A Heuristic Wolf Pack Algorithm （HWPA） is proposed to solve the unrelated parallel machine scheduling problem aiming at minimizing the maximum completion time. Firstly， considering the search efficiency of the WPA and the computational resource usage， a new heuristic algorithm is established to generate the initial population. In order to improve the efficiency of the algorithm， the makespan was used to set the corresponding local search mechanism， which was optimized at the same time with the randomly generated initial population； Secondly， the intelligent rules are designed according to the characteristics of the problem and the walking around mechanism in the algorithm is adjusted to replace the fixed step size with random step size. Differentiated local optimal strategies are adopted in the summoning and besieging mechanism to jump out of the local optimal solution and improve the search efficiency of the algorithm； Finally， the new algorithm is used in simulation and optimization experiments. Experimental results show that the algorithm is 107.35% better than GA and 113.62% better than FOA in less iterations. The effectiveness and superiority of the algorithm are proved.关键词：parallel machine scheduling;heuristic wolf pack algorithm;intelligent behavior mechanism;local search mechanism;local optimization strategy
- 14
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：Obtaining the accurate estimation of indoor occupant counting in different areas is of great significance in many applications， including building energy conservation， intelligent regulation of air conditioning and lighting systems， and public security. Therefore，propose a low-cost， high-accuracy Wi-Fi-based indoor personnel identification general model. This model deploys multiple Wi-Fi probes， collects terminal sample data in different environments in the offline phase， uses particle swarm optimization to optimize the BP neural network， and uses a parallel method to quickly train the personnel identification model； in the online stage， based on the trained personnel identification model， judge whether the personnel are in the area according to the actual received Wi-Fi signal strength， and finally count the number of people in the area. Finally， in a research building of Anhui Jianzhu University， a Wi-Fi-based indoor people identification system was built， and the effectiveness of the method was tested through experiments. The results show that the personnel identification model in all areas involved in the experiment， the recall rate of the model is greater than 90%， and the model accuracy rate can reach more than 91%，which meets the practical application requirements of building intelligent system for the rapid deployment of personnel quantity measurement method.关键词：Wi-Fi sniffer;personnel identification;BP neural network;particle swarm optimization;parallel processing
- 18
- |
- 7
- |
- 0
发布时间：2022-08-18 - 摘要：Aiming at the problems of slow convergence speed and integral Windup effect in the traditional integral sliding mode control of permanent magnet synchronous motors（PMSM）， an new integral fast terminal sliding mode control strategy was proposed. First of all， this control strategy combined traditional integral sliding mode and fast terminal sliding mode control methods to achieve fast global convergence of state variables. Then， the integral term in the sliding mode surface was redesigned into the integral form of nonlinear function and fuzzy value， which can effectively avoid the integral Windup effect while eliminating the steady-state error. Finally， the control method was applied to the speed controller of the permanent magnet synchronous motor vector control system， and compared with some control methods such as traditional integral sliding mode control method. The experimental results show that the controller can effectively improve the dynamic and static performance of the speed regulation system. The strategy avoids the problems of traditional integral sliding mode and provides an effective method to improve the dynamic quality of PMSM speed control system.关键词：PMSM;terminal sliding mode;nonlinear function;fuzzy control
- 14
- |
- 5
- |
- 0
发布时间：2022-08-18 - 摘要：ArcGIS was the mainstream GIS application and development platform at present， but the procedure for license installation and setting methods was quite complex. In order to reduce the difficulty of installing software for non-GIS professionals and reduce the workload of configuring licenses for multiple computers in laboratories， it is necessary to deeply study the operating mechanism of ArcGIS licensing services. Based on the characteristics of the Windows operating system and the ArcGIS operation method， it is found that ArcGIS license configuration information was stored in the registry table. Further exploration revealed the data item details in registry table and operation principles of ArcGIS for Desktop， License Manager Administrator and ArcGIS Administrator. Based on this theory， an automatic license setting APP was developed.In the past， it took half an hour to an hour to manually configure licenses on a computer， and now it only took 1 minute by using this APP. The APP solved the problems of automatic， batch licensing and automatic adaptation of different versions of ArcGIS.关键词：Arcgis;geographical information systems;license configuration
- 28
- |
- 5
- |
- 0
发布时间：2022-08-18 - 摘要：In order to improve the success rate of traditional tachyarrhythmia ablation procedure， shorten the operation time and strengthen clinical training， design a new kind of data transmission scheme and 3D tracking solution， using HoloLens2， MRTK API， ZeroMQ and magnetic positioning and Vuforia hybrid tracking technology， developed EP （Electrical Physiology） navigation system. The model display effect test， data transmission performance test and 3D tracking registration accuracy test proved the usability of the operating navigation system. The experimental results show that when the number of triangles of the model is less than 15k， the model shows that the frame rate is mainly distributed between 50~60FPS， accounting for 78% of the total， and real-time data transmission can ensure that the data is not lost and the overall delay is less than 1ms. Under the circumstance of ensuring the effect and safety of the operation， the influence of various factors （identification map position， identification map size， observation angle， observation distance and catheter movement speed） is minimized， and the best practice plan is determined， with a minimum error of 1.17 mm.关键词：augmented reality;tachyarrhythmia;EP navigation system;HoloLens2;data transmission;hybrid tracking registration
- 23
- |
- 9
- |
- 0
发布时间：2022-08-18 - 摘要：At present， X-band Doppler weather radar location geographic information data acquisition is difficult， processing is complex， there is data redundancy or data loss， and it is impossible to accurately evaluate the detection capability of X-band Doppler weather radar. Therefore， a convenient and refined location selection method for X-band Doppler weather radar is proposed by combining the GSCloud platform and LSV software. The high-precision geographic information data downloaded from the GSCloud platform are combined with LSV software to process the data pertinently， so as to produce the geographic information data corresponding to the working area of the proposed radar station. The practical results show that the proposed method has high accuracy and strong operability， and can better assist the location of X-band Doppler weather radar to achieve the maximum detection capability. The proposed method provides some reference for the location of X-band Doppler weather radar.关键词：X-band;radar location;geographic information;GSCloud;LSV
- 20
- |
- 7
- |
- 0
发布时间：2022-08-18 - 摘要：The cigarette marketing system is usually very large and complex， and the cigarette marketing data in this system are from many sources which include not only the cigarette marketing system itself but also the third-party systems. These data characters always bring great challenge to the bloodline analysis of cigarette marketing data. In this work， we design a fine-grained bloodline analysis method of cigarette marketing data based on the proxy re-signature protocol. Comparing with the existing solutions of data bloodline analysis， the proposed method can support fine-grained bloodline analysis and ensure the data bloodline is not tampered in the cloud outsourcing service scenario. The experimental results validate that the proposed method satisfies all the requirements of bloodline analysis of cigarette marketing data. By the proposed algorithm， the query efficiency and the verification efficiency of bloodline analysis of cigarette data reach 1，023 milliseconds and 970 milliseconds. which are able to provide efficient， secure and fine-grained bloodline analysis services for cigarette marketing data under the scenario of cloud data center.关键词：bloodline analysis;proxy re-signature;cigarette marketing data
- 49
- |
- 11
- |
- 0
发布时间：2022-08-18 - 摘要：In the Internet environment， network public welfare is developing rapidly. It is particularly important to analyze the pattern of information dissemination of public welfare. Based on the information ecology theory， this paper decomposes the network public welfare operation mechanism from the three dimensions of information， information person and information environment， and constructs the network public welfare information ecosystem model. Taking the "blue lifeline" topic of sina Weibo public welfare as the research object， combined with the network characteristic quantity， this paper analyzes the characteristics of public welfare information， information person and information environment under the network environment， and obtains the characteristics of network public welfare information and the pattern of user behavior. The results show that the characteristic quantity of information person conforms to the piecewise power-law distribution. In the information ecological chain， the information producers and information organizers have strong information influence， and only a few information disseminators play the role of secondary communication； The transformation of public welfare information into power-law distribution shows that most information has not been spread， and only a few information has been received by the public； The higher the popularity of public welfare information environment， the more users participate in public welfare. Therefore， we should pay attention to the publicity of public welfare activities， the quality of public welfare information and the cultivation of the influence of public welfare organizers and producers.关键词：online public welfare;information ecology theory;network feature;user behavior
- 14
- |
- 4
- |
- 0
发布时间：2022-08-18 - 摘要：Aiming at the problems of high coupling between functional modules， poor concurrency and poor data query performance of traditional shopping website system， a shopping website system based on microservice technology is constructed. By constructing a framework based on Dubbo and integrating SSM （SpringBoot-SpringMVC-Mybatis）， the coupling between the modules of the shopping mall is reduced； Add Redis cache database to improve data reading and writing performance， forward service requests to different service nodes for business processing through the polling strategy in Nginx load balancing technology， and improve the concurrency and availability of the system. Using PostMan and JMeter tools for performance testing， the results show that the average response time of the shopping website system using Redis database and Nginx load balancing technology has been shortened by 28%， and the throughput performance has increased by an average of 43%.关键词：shopping website;Redis cache database;Nginx load balancing technology;performance optimization
- 25
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：Aiming at the problems of the current traditional Bloom filter element deletion difficulty and the difficulty of eliminating the false judgment rate， a new type of erasable high-efficiency filter algorithm based on bit identification is proposed. The algorithm uses an improved prefix tree to construct an erasable high-efficiency filter， and uses its structural characteristics to solve the problem of difficult element deletion in the traditional Bloom filter and achieve zero misjudgment rate. According to the performance optimization strategy， the traditional R-direction prefix tree is improved based on the bit mark， which greatly reduces the memory consumption. Experimental results show that this algorithm can efficiently complete the retrieval and filtering of strings， reduce the consumption of memory space under the premise of ensuring time complexity， and can delete filter elements to achieve zero false positive rate， which is suitable for high concurrency scenarios system applications.关键词：bit identification;prefix tree;bloom filter;erasable filter;string search
- 10
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：In order to solve the problem of data integration in the process of natural resources information system construction， realize the business data conversion and synchronization of natural resources departments more efficiently， safely and flexibly. Based on the business requirements， analyze the current application status of data integration technology， carries out the research on the open source ETL tool Kettle， analyzes the Kettle conceptual model and its application scenarios， combines with the characteristics of wide data sources， large amount of data and complex data structure in the construction of natural resources information system. A Kettle-based source database synchronization environment is constructed， and a Kettle-based data transformation synchronization method is proposed. At the same time， this method is applied to the actual project case， compared with the traditional method， this method improves the development speed and work efficiency of ETL， not only solves the problem of conversion and synchronization from multi-source data to target data in the process of natural resources information construction， but also provides more ideas for other enterprises' data integration work.关键词：natural resources informatization;data synchronization;Kettle;ETL
- 21
- |
- 3
- |
- 0
发布时间：2022-08-18 - 摘要：Aiming at the problem that the image is easy to distort in the process of insulator detection and the model is greatly affected by the external environment， an insulator detection algorithm based on YOLOv4 is proposed. Firstly， adaptive gamma transform is used to automatically adjust the brightness of the insulator image. Then， the YOLOv4 network is used to learn the feature representation at different levels of the insulators. Considering that the YOLOv4 network needs to input a fixed image size， and the forced stretching of the image will distort the target， deformable convolution is adopted to replace the traditional convolution method， so as to improve the feature extraction ability of the model. Finally， the final experimental results are obtained after the insulator position information and its categories are output. The experimental simulation on the public database of power insulators in China shows that the test accuracy and detection speed of the proposed algorithm are 93.2% and 43FPS， respectively.The overall performance of the algorithm is better than Faster RCNN， YOLOv3， CornerNet and other common algorithms.关键词：insulator;image distortion;deformable convolution;YOLOv4
- 29
- |
- 5
- |
- 0
发布时间：2022-08-18 - 摘要：Aiming at the traditional Census transform that overly relies on the central pixel and the time-consuming calculation of its cost increases with the increase of the support window， an improved AD-Census transform is proposed to be applied to the binocular ranging system. The improved AD-Census transform selects 8 equidistant pixels around the center pixel for pairwise comparison to obtain a bit string of one byte， and XOR the corresponding bit string in the left and right fields of view to obtain the Hamming distance as the initial of the corresponding window Cost， get the corresponding disparity map， and finally calculate the distance of the target. From the time-consuming comparison experiment， it can be seen that the calculation time consumption of the algorithm in this paper is stable at about 0.2s and 6.8s for low-pixel graphics and high-pixel graphics， which is nearly 1/3 of the time for traditional AD-Census conversion. In the distance measurement experiment， when the distance between the measured object and the optical center of the camera is measured from 800mm to 3000mm using the algorithm proposed， the absolute value of the relative error percentage between the measured and actual distance is less than 5%. The algorithm in this paper not only greatly improves the stereo matching speed， but also the accuracy of the algorithm meets the requirements of the ranging experiment.关键词：computer vision;AD-census conversion;stereo matching;binocular ranging
- 14
- |
- 4
- |
- 0
发布时间：2022-08-18 - 摘要：The purpose of brand clothing image retrieval is to directly identify the clothing that is similar to the target image according to the user's requirements. The process of apparel image retrieval consists of two steps. Firstly， the image features of the dataset were extracted by the convolution neural network model and saved to the database， then the features of the image to be retrieved were extracted and the similarity between the features of the image to be retrieved and the saved image features was matched. Improve the VGG16 model by pruning the convolutional kernel from the design of deep convolutional neural network model， analyzed the structure of Inception_v3 and the original VGG16 model， and verified its performance through experiments. The experiments show that the accuracy of the improved modify_vgg model is 81.7% on the brand clothing dataset， which is 1.4% higher than the original VGG16 model， and 3.5% higher than that of the Inception_v3 model.关键词：convolutional neural network;image retrieval;feature extraction;VGG16;Inception_v3
- 28
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：In order to solve the problems of hiding capacity and hidden image quality of existing reversible data hiding algorithms， a reversible information hiding algorithm based on univariate quadratic equation interpolation space is proposed. According to the gradient prediction property， the interpolated image block is divided into edge block and non edge block， and the block with small variance is preferentially selected to embed the information， and the univariate quadratic equation interpolation algorithm OQEIAM and adaptive gradient prediction interpolation algorithm AGPIA are used to hide the non edge block and edge block information. The test results on SIPI dataset show that the algorithm has a full load average capacity ER of 4.068 1bpp， which is higher than most similar algorithms， and has high image quality， without additional information and data overflow in the process. The proposed method significantly improves the embedding rate and image quality in large capacity reversible information hiding， and provides a new idea for this field.关键词：reversible information hiding;one-variate quadratic equation interpolation;adaptive gradient interpolation;large capacity;high image quality
- 22
- |
- 7
- |
- 0
发布时间：2022-08-18 - 摘要：Brain tumors are abnormal cell growths in the brain and are highly dangerous diseases due to cancerous changes in the internal tissues of the skull and brain. At present， the diagnosis of brain tumors is very dependent on medical imaging technology， among which magnetic resonance imaging is the most widely used. Brain tumor segmentation based on MRI images is of great importance. In this paper， we propose a brain tumor image segmentation network based on U-Net improvement， combining a residual network and a module for enhancing contextual information， and adding a null space convolutional pooling pyramid processing. The experimental validation of the MRI image dataset of brain glioma provided by the TCIA shows that the correct rate of the brain tumor image segmentation network based on U-Net improvement proposed in this paper reaches 0.957 2， which can effectively improve the segmentation accuracy and enhance the efficiency of brain tumor recognition， and has positive significance for the clinical diagnosis of brain tumor.关键词：neural network;brain tumor;medical image segmentation;U-Net
- 23
- |
- 17
- |
- 0
发布时间：2022-08-18 - 摘要：The deep meaning of mural culture is obscure due to its inheritance and dating，the dynamic digitization of mural content based on virtual animation technology provides new ideas for the protection and inheritance of traditional mural culture. Specifically， the gray level distribution is obtained through the original image， and the gray level distribution is adjusted according to the black-and-white field threshold and intermediate tone， so as to improve the contrast and brightness of the image， enhance the edge details of the foreground and background， and extract the primitive. Through the triangular mesh features of primitive characters， the two-dimensional bone relationship is established to complete the binding and weight setting. Based on the skin matrix and weight factor， the coordinate mapping between bone and mesh vertices is completed and the bone matrix is generated. The relative coordinates of the vertices in the bone matrix are analyzed according to the time axis， and the world coordinates of the vertices are calculated through the coordinate mapping relationship to complete the action description of the primitive character. Relying on the in-depth analysis of mural culture and virtual animation technology， this paper studies the interaction from image to graphic element， from single graphic element to multi graphic element， and completes the research on the static digitization of mural graphic element and the dynamic digitization of plot animation interaction.关键词：mural elements;element extraction;bone drive;plot animation
- 20
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：In order to solve the problems of deaf-mute's communication barriers in daily life and work and that traditional gesture recognition methods are easily affected by complex environment， a static gesture recognition algorithm based on the combination of YCbCr color space and convolutional neural network was proposed. The algorithm firstly extracted the hand region based on YCbCr color space， Then， the extracted images are preprocessed with dimensionality reduction， gray scale and data enhancement， and the convolutional neural network model is used to train and classify the preprocessed images. Finally， experimental verification shows that the recognition accuracy of this algorithm on NUS-II and Marcel complex background gesture data sets reaches 98.32% and 98.96% respectively. Compared with traditional algorithms and other algorithms based on convolutional neural networks， the recognition rate is higher and the recognition effect is better.关键词：image preprocessing;complex background;image enhancement;convolutional neural network
- 33
- |
- 12
- |
- 0
发布时间：2022-08-18 - 摘要：The shape， size and color of license plates of different types of vehicles vary. In order to accurately detect license plates under different shooting views， scales， backgrounds， light intensity and various forms of occlusion， a multi-stage license plate localization model based on SURF algorithm is proposed. The model considered the rich texture and structure information of the license plate， using the row covariance coefficient distribution of the SURF feature matrix to define the characteristics of the license plate row candidate area， so as to obtain multiple block areas with obvious differences， and propose a new four-dimensional feature descriptor for precise extraction of the candidate area of the license plate row， finally， the identification of the license plate area is realized based on the Hessian matrix to measure the structural characteristics of the license plate characters. After testing on the CCPD data set， the model does not require any controlled conditions or environmental settings， and has the ability to deal with license plate deformation， blurring， dirtying， and lighting changes.关键词：SURF algorithm;license plate location;feature descriptor;computer vision;intelligent transportation
- 16
- |
- 3
- |
- 0
发布时间：2022-08-18 - 摘要：In view of the difficulties faced by the current teaching experiment conditions， funds， teaching methods and management methods， a virtual simulation experiment teaching network platform is studied and established， so that the funds for teaching experiments can be shared resources， but also to enable students to practice and innovation awareness has been promoted. Taking the construction of computer virtual simulation experiment teaching platform in Inner Mongolia University as an example， the process of constructing virtual simulation experiment teaching resources is discussed， and the mode of sharing teaching experiment resources is introduced， then through the application of the platform to show the practical effect.The practical results show that by using the experimental teaching platform， students' academic performance has been improved， and their practical ability and innovation ability have also been improved.关键词：virtual simulation;computer;experimental teaching platform;teaching reform
- 132
- |
- 26
- |
- 0
发布时间：2022-08-18 - 摘要：According to the "Shuangwan Plan" for building first-class undergraduate professional points issued by the Ministry of Education， based on the high-level， innovative and challenging standards of gold course construction， and taking the high-level language programming course as the research object， gold course construction is carried out from the aspects of the main contents of course resource database construction， hierarchical construction objectives， construction team and cooperation division of labor， hierarchical topic database and project database construction， course assessment and evaluation， etc. The practice results show that the overall performance of students shows an upward trend， and the innovation and application ability is also improved.关键词：high-level language programming;golden courses;Shuangwan Plan;resource database
- 26
- |
- 8
- |
- 0
发布时间：2022-08-18 - 摘要：According to the curriculum characteristic and teaching content of introduction to artificial intelligence， aiming at both improving the interest and stimulating the learning initiative of students， combined with teaching practice experience of the group members in artificial intelligence and its relevant courses， discuss the teaching reform scheme for introduction to artificial intelligence course： from tracking the development of military intelligence， multi-channel learning by online and offline， organizing the teaching content according to the clue of example-conception-application， mastering the basic principles and methods in practice， cultivating students' innovative consciousness and forward-looking consciousness， strengthening the construction of experimental environment and teaching environment in detail. Practice shows that this method can better stimulate the learning interest of students and achieve good teaching results.关键词：Introduction to Artificial Intelligence;teaching methods;military intelligence;online and offline teaching
- 28
- |
- 11
- |
- 0
发布时间：2022-08-18 - 摘要：Considering the education of intelligent software analysis， we have systematically investigated the courses of first-class universities in the world. As a result， we propose a three-level course content for teaching the principles of intelligent software analysis： underlying theories， key techniques， and classical applications. Besides， we have also carried out some course teaching and assessment reforms. These reforms achieve a good teaching result.关键词：intelligent software analysis;course design;trustworthy software
- 7
- |
- 4
- |
- 0
发布时间：2022-08-18 - 摘要：In military academies， how to carry out the innovative education and teaching of "artificial intelligence" + "second classroom"， and expand the students' innovative thinking and practical ability is an urgent problem to be solved in the cultivation of innovative talents. Following the teaching idea of "learning oriented" and "multiple subjects" as the characteristics， the teaching reform and exploration of intelligent robot "second classroom" are carried out from the aspects of teaching design optimization， teaching case enrichment and teaching method renewal. The reform can effectively stimulate students' learning enthusiasm， strengthen teachers' professional ability， and improve the level of professional construction.关键词：intelligent robot;multiple subjects;Innovative Practice;teaching reform
- 9
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：In view of the problems in the course of ideological and political implementation， such as the lack of ideas and the lack of ability， and combining with the characteristics of “computer system and network”， such as the course is practical and has a wide range of influence， combing the basic goal and general train of thought of the ideological and political construction of the curriculum， expound the concrete implementation process of the ideological and political construction of the curriculum from six aspects of improving the ability of educating people， mining the teaching content， revising the teaching plan， innovating the teaching means， building the comprehensive system， and constructing the dynamic evaluation system. The results of the implementation of different classes in two semesters of the course show that， the implementation of the program of ideological and political education can improve the effect of education remarkably.关键词：computer system and network;ideological and political construction;ideological and political education;military academy
- 16
- |
- 6
- |
- 0
发布时间：2022-08-18 - 摘要：With the rapid development of digital technology and the Internet， information security is facing more and more challenges， and text steganaography， as an important technology in the field of information security， has become a current research focus. This article is based on 200 articles collected in core database of web of science in the past decade， CiteSpace visualization software is used to analyze knowledge base， development track and research focus of text steganography from on the number of published papers，countries distribution and the cooperation institutions，co-occurrence of key words and co-citation of literature knowledge and so on. The results showed that： ①The amount of literatures of text steganography showed a trend of steadily increasing started 1995， especially since 2016，the number of literatures showed a sharp growth；②In the network of institutional cooperation， the core research institute is Nanjing University of Information Science & Technology. The connection among the institutes is generally weak；③The research of text information hiding is evolving in solving the problems of concealment and hiding capacity. In recent years， the research focus has switched to coverless text steganaography， as well as the corresponding steganalysis and hiding capacity. The research results provide reference information for researchers in the field of text information hiding， hoping to achieve the purpose of reducing the subjectivity of research.关键词：text steganaography;steganalysis;CiteSpace;hiding capacity;hiding security
- 24
- |
- 13
- |
- 0
发布时间：2022-08-18 - 摘要：Cephalometric analysis is an indispensable clinical and research tool in orthodontic analysis and treatment planning. The marking of cephalometric marker points is an indispensable step in cephalometric analysis. With the development of image processing technology， the marking of cephalometric marker points has also undergone a process from manual marking to automatic marking. In recent years， more and more automatic marking methods for cephalometric marker points have been proposed， ranging from image processing methods to statistical template matching methods， to current machine learning methods and deep learning methods. Systematically review the application progress of automatic recognition of several cephalometric marker points， and further discusses its existing problems and research directions， in order to provide a certain reference and reference for the research in this field.关键词：Cephalometric analysis;auto recognition;image processing;machine learning;deep learning
- 24
- |
- 12
- |
- 0
发布时间：2022-08-18 - 摘要：As a hot direction in the natural language processing field， Text automatic proofreading has been widely studied. According to the different types of Chinese text errors， it can be divided into three directions： spelling error correction， grammar error correction and semantic error correction. Firstly give a brief introduction to text proofreading； Secondly， the Chinese text proofreading models using traditional methods and deep learning methods are analyzed and summarized respectively， points out the problems existing in this field， and puts forward improvement schemes.Through the research and analysis of the current Chinese text automatic proofreading methods， in order to provide reference for scholars in this field.关键词：Chinese text proofreading;natural language processing;language model;deep learning
- 41
- |
- 16
- |
- 0
发布时间：2022-08-18 - 摘要：With the development of science and technology， technology has driven the constant innovation of education. It has not only changed the structural elements of education，promoted the integration and development of technology and education， but has also set off a wave of education evaluation reform. As the focus of educational modernization， the smart educational evaluation is an important direction of education evaluation reform.By analyzes the development status of smart educational evaluation， and believes that the current smart educational evaluation in China is still in the exploratory stage，with limitations in conceptual focus， basic conditions， and technical support. In terms of the future development trend of educational smart evaluation， it is suggested to strengthen the concept attention， attach importance to the individuation， process and holography of evaluation， optimize the basic conditions， improve information literacy， promote multi-field cooperation， enhance technical support， realize high-performance data processing， and improve the evaluation model and system. At the same time， it clarifies the applicability of evaluation， avoids ethical privacy and excessive technical bias， so as to realize the healthy development of educational wisdom evaluation.关键词：smart evaluation;smart education;education evaluation;evaluation technique
- 14
- |
- 9
- |
- 0
发布时间：2022-08-18 - 摘要：Atrial fibrillation is a serious disorder of atrial electrical activity， which can lead to heart failure and stroke. With the development of artificial intelligence， deep learning algorithm is more and more widely used in the early detection of atrial fibrillation. CiteSpace software was used to retrieve 174 relevant literatures collected by CNKI from 2001 to 2022， and bibliometric and visual analysis was carried out from the four aspects of publication year， author， organization and keywords， so as to present the research trend in this field. The results show that the number of papers published in the relevant literature is on the rise， but the total amount is still low； The degree of cooperation between authors and institutions is low， and the research distribution is sparse； The research mainly focuses on transfinite learning machine， R wave localization， Greek bundle， pattern recognition， QRS wave detection， ventricular fibrillation， P wave， blind source extraction， independent component analysis， machine learning， preprocessing， chaos， information entropy， wave graph， statistical performance， etc. Using deep learning algorithm to detect atrial fibrillation has become a trend， but there is still much room for improvement in domestic research in this field， such as the research needs to be diversified， the algorithm fusion needs to be promoted， and the communication and cooperation between research teams and institutions need to be further strengthened.关键词：atrial fibrillation detection;CiteSpace software;deep learning
- 15
- |
- 7
- |
- 0
发布时间：2022-08-18 - 摘要：With the continuous development of blockchain technology， the application of blockchain in human resources and social security business is becoming more and more extensive. As an emerging technology， blockchain has the characteristics of decentralization， immutability， traceability， and high security， which can improve the online services of human resources and social security business， and meet the needs of data sharing， collaborative services， and supervision. However， the application of blockchain technology in human resources and social security business has just begun， and whether it can meet technical requirements such as unified planning and construction， data storage and compatibility， and reasonable selection of business scenarios remains to be further explored. First conduct relevant research on the application of blockchain technology in the human society business， analyzes the current problems in the human society business， and makes a preliminary exploration of the application of blockchain technology in a specific scenario of human society business. Then analyze the challenges faced by the deep integration of blockchain and human and social business， and look forward to the development prospects of blockchain technology in human and social business applications， to provide reference for relevant departments.关键词：blockchain technology;blockchain+human resources and social security;consensus mechanism;smart contract
- 14
- |
- 8
- |
- 0
发布时间：2022-08-18
- 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