Abstract:The rapid development of the Internet has led to increasing online shopping and accumulated large amount of complicated, multi-dimensional and temporal transaction data. Many e-retailers and analysts are focusing on these data to study customers' shopping behaviors. In this study, based on product gravity, a visual analytical method focusing on individuals is proposed to analyze hierarchic production data and the features of shopping behaviors, which allows analysts to observe each consumer's shopping behaviors more directly so as to discover the features of their shopping behaviors more easily. Finally, the experimental results are presented to prove the effectiveness of the method.

Abstract:This study focuses on emotion recognition technology. The input signals are EEG and facial expression. The stimuli are based on a subset of movie clips that correspond to four specific areas of valance-arousal emotional space. In facial expression detection, one of the four basic emotional states is determined. In EEG detection, one of the three emotional intensities is determined. Emotion recognition is based on a decision-level fusion of both EEG and facial expression detection. The results show that the accuracy of information fusion detection is 89.5%, which is higher than that of facial expression (81.35%) or EEG detection (71.53%).

Abstract:In order to study the interaction between different systems in the workflow process, the interaction polychromatic graph and interaction matrix of workflow process are constructed based on the polychromatic set theory. Firstly, we combine the tuples and polychromatic set theory to form multi-element binary polychromatic sets and single-element binary polychromatic sets. Secondly, we expand the contour matrix, constituting the K-tuple contour matrix and its disjunction, conjunctive, connected operations, and zone the nodes in the polychromatic graph to represent the resources and messages required for the interaction between the activities. Finally, we construct the interaction model between the system and the workflow process, with the interaction between the hospital HIS and RIS workflow as an example for verification.

Abstract:Face verification is important for personal identity authentication, which is significant in system security and criminal identification. Face verification task is to give a pair of face images to determine whether they are of the same identity (i.e. binary classification). The traditional authentication method consists of two steps: feature extraction and face verification. In this study, a hybrid convolutional neural network (HBCNN) is proposed for face verification. The main process is divided into three steps: feature extraction, feature selection, and face verification. The key point of this model is to directly use the mixed convolutional neural network to learn the relevant visual features directly from the original pixels and to further process the features through univariate feature selection and principal component analysis (PCA). This can be achieved from the original pixel extraction to a better robustness and expression of the characteristics. The support vector machine (SVM) at the top level is used to see if it is the same person. Experiments show that the mixed convolutional neural network model has a better performance than the traditional method in verifying accuracy of face verification.

Abstract:A table data simulation generating algorithm is proposed based on not-temporal attribute correlation. This algorithm can overcome the difficulty in building not-temporal attribute correlation in the development of big data simulation generator, and play an important role in the field of measurement of the big data simulation generated. Firstly, we extract the two key not-temporal attributes from the data set, and make the statistics of twofold frequency. Then, based on the statistical results, we calculate the maximal information coefficient (MIC) value to measure dependence for two-variable relationships. We use the stretched exponential (SE) distribution to fit the relationship, and build the correlation model. Finally, we generate data in a two-dimensional matrix with this model. The experimental results show that this algorithm can effectively describe the data characteristics of the real data set.

Abstract:This study designs and realizes the solar-thermal experiment system based on STM32. It introduces the structure and function of the system as well as the design of hardware and software of the system terminal. It analyzes the operation principles and circuit structures of the functional circuits in the solar-thermal experiment system, which consists of the temperature sensor PT1000 modulation circuit, AD conversion and isolation circuit and artificial light dimmer circuit unit. The paper proposes an architecture multitasking software system, using the least squares fitting method for piecewise linearization of the measured temperature data, and introduces the error evaluation principle to determine the best suitable for PT1000 correction equation. The related technology has been applied to solar-thermal experimental teaching system of New Energy Engineering Center of Fujian Normal University. The application shows that the terminal performance indicators meet the design requirements.

Abstract:In this study, the DDS (Data Distribution Service) technology of OMG (Object Management Organization) is applied to the command and control system to achieve the system's loose coupling in time, space, data-flow so as to reduce the system's coupling degree and the difficulty of integration. At the same time, based on the analysis of the shortages of the existing commercial DDS software that is compied according to the DDS standard, this study puts forward the application of DDS to the command and control system. Furthermore, it adopts QoS (Quality of Service) control, bandwidth adaptive, scalable binary variable length coding, multi-priority queues, technologies of reliable information transmission and other technical methods to further improve the efficiency and reliability of the data transmission in the command and control system. Then, the synthesized effect that DDS and related technical methods is applied to the command and the control system is compared with the technical indexes of the existing command and control system. The results show that the application of DDS to the command and control system can effectively solve these issues in the existing command and control system.

Abstract:Cloud platform monitoring system is an important segment to effectively guarantee the quality of cloud services. In this study, the corresponding virtual server monitoring system is designed on the Openstack cloud platform. The system combines the characteristics of Openstack's open source and strong scalability. At the same time, it starts with the bottleneck problem which appears when Neutron, the network component in Openstack faces the huge virtual server traffic forwarding. This study combines SDN technology and openflow, its important southward interface protocol, to solve the traffic bottleneck problem. At the same time, it designs the monitoring system based on Libvirt and sFlow protocol responsible for obtaining data flow of virtual cluster, giving feedbacks of the latest platform load to the upper control module and user, and using control module and flow table to control traffic forwarding. Finally, the whole system platform can achieve load balancing.

Abstract:In seismic data processing environment, the seismic data writing has imbalance without seismic data distribution optimization in the file system pool when large scale seismic data frequently perform writing and deleting operations. This study analyzes the seismic data processing classification and various conventional writing strategies of seismic data. It proposes a seismic data writing strategy based on the idea of probability which meets the optimization distribution of seismic data writing under different initial files in system pool. The experimental results show that the method is effective.

Abstract:In recent years, with the development of informationization construction of enterprises and institutions, voting appraisal becomes an important method for appraisal decision. However, the traditional voting system has widespread poor stability and scalability, and it is difficult to cope with various unexpected situations in the voting process. In this study, a real-time and high reliable voting system is realized by using the finite state machine and WebSocket technology. The system has high scalability, and can be flexibly configured to realize a variety of forms of voting information management, and can expand a variety of templates to meet the growing business needs. Several practical uses have proved that the system realizes a rapid and stable voting process, and has greatly improved the appraisal efficiency.

Abstract:In view of the loopholes and defects in the application softwares of Android mobile phones, this paper analyzes the current solutions of mobile phone login system, using multi factor (account number, password and verification code, login numbers, login location, face recognition) scheme to build secure login system for mobile phones. The login system consists of login, registration, login audit, WeChat alerts, and it introduces the design ideology and technology route. It mainly introduces security verification logic and log audit function, the user identification and login behavior audit, to provide users with a safe, easy-to-use, and low cost solution.

Abstract:With the widespread use of WeChat, more and more enterprises begin to make the secondary development through the Wechat public platform. In order to allow property buyers to enjoy safe and efficient supervision services and rapid and high-quality advisory services, to ensure that regulatory authorities and property buyers can jointly supervise the commercial housing pre-sale funds, this study builds a bridge for efficient communication between regulatory authorities and property buyers, to avoid the developers' mistakes in failing to report the amount of the funds or making a wrong report. A WeChat supervision service platform is hereby designed and implemented for commercial housing pre-sale funds. Message push and real-time capital supervision are implemented on this platform. By making full use of the data in the database of the information sharing platform of commercial housing pre-sale funds supervision, the content of the message will be enriched, and the supervisors' and property buyers' information inquiry requirements will be met.

Abstract:In order to accurately depict the randomness of the node behavior of the P2P file-sharing systems, a dynamic model of the P2P file-sharing systems based on the online-probability is proposed. Firstly, we introduce the online-probability of the nodes and analyze the process of the evolution of the systems. Furthermore, we propose a dynamic model of P2P file-sharing systems based on the online-probability. Through the model, the factors that influence the system are studied. These factors are embodied in the central policies of the system. The relevant algorithms of the dynamic model of the P2P file-sharing systems is improved. The relevant algorithms based on the online-probability are proposed. These policies include peer selection policy, bandwidth allocation policy and peer choking policy. Finally, some experiments are carried out to validate the model.

Abstract:This paper presents the studies and implementations of river and lake information system based on WebGIS technology in the Tarim river basin, which is concerned with the low efficiency of information acquisition of rivers and lakes in the Tarim river basin. The system mainly classifies the natural information, humanistic information, and spatial information of the rivers and lakes in the nine rivers in the Tarim river basin, Xinjiang, and realizes the display operation of the spatial data, the search and positioning of the rivers and lakes, multimodal query of river and lake information, rivers and lakes of the calculation, and other functions. The test results show that the system provides a convenient way to obtain information in the Tarim river basin, and has good practicability.

Abstract:To deal with the high cost and the inflexible opening hours for traditional labs, a 3D virtual laboratory system based on X3DOM engine has been designed. The system is developed based on the browser/server architecture. The server provides laboratory scene, virtual experimental instruments, XML template, and file storage and access services. The browser uses MVC design pattern. According to the idea of component development, X3DOM is used to render the view. Javascript is used to build the simulation module and the extensible component library is built. Taking the Ping experiment in the computer network course as an example, the design and implementation of the virtual laboratory has been verified. The results show that the system can be used to effectively assist experimental teaching.

Abstract:Aiming at the daily security of students in colleges and universities, in this study, we proposed and designed a set of security management system for college students based on ThinkPHP framework and Workerman framework, focusing on the whole process from the data receiving, the data storage, and ultimately to the realization of data display and its application. According to the principle of symmetric encryption algorithm, we set the encryption rules independently, improving the system security. The system has the advantages of clear power design, efficient code operation, and high safety. With this system, administrators can query the students' basic information, current real-time location, daily attendance record, and alarm data, which has effectively solved the problem of the safety management of students in colleges and universities.

Abstract:The performance of min-sum (MS) decoding algorithm depends on each bit corresponding to the accuracy of log-likelihood ratio (LLR) for multi-level cell (MLC). However, uniform sensing strategy needs to increase the sensing precision in order to obtain high accuracy of LLR, which increases the reading latency of MLC. In this study, an MS decoding algorithm is proposed for MLC flash memory, which uses non-uniform sensing strategy for the threshold voltage of MLC. In the same sensing precision, compared to the uniform sensing strategy, the non-uniform sensing strategy can improve the accuracy of LLR and lower the raw bit error rate. The simulation results show that the proposed method can not only guarantee the reliability of the MLC flash memory, but also keep the fast reading speed for MLC flash memory, thus achieving a better tradeoff between complexity and decoding performance.

Abstract:In the IOT of intelligent agricultural greenhouse system, the data collected by the sensor nodes need to be uploaded to the remote server via the Internet, and 6LoWPAN is used between the Contiki nodes. It cannot connect directly with the traditional IPv4, so it is necessary to configure an Internet of Things gateway. By transplanting the OpenWRT system to HG255D router, the router can be changed into a small, practical embedded Linux system, so that they can run open source or developed program. This article introduces how to implement HG255D routers on the IPv6 Internet access gateway, mainly involving the compilation of the OpenWRT firmware, router configuration changes, the transplantation of gateway program, as well as the effect of demonstration.

Abstract:With the development of integrated and modularized civil aircraft electronic system, IMA (Integrated Modular Avionics) system comes out. It manages and deploys many individual UAs (User Applications). CDS (Cockpit Display System) which is in the civil avionics system focusing on IMA, provides the display function for UA. At the same time, CDS manages the IDUs (Integrated Display Unit) which provides the HMI for cockpit giving the feedback to UA. Developing the UA code and ARINC 661 protocol DF (Definition File) which is loaded on IDU are the responsibility of UA developers. But they are not implemented in one target PC. So, they are individual and related. In order to according with DO-178C and CAAC's certification, we do the research of development and verification of DF.

Abstract:In view of the vehicle's abnormal behaviors in the artificial monitoring, such as speeding, illegal lane changing and red light running, this study proposes a method for detecting abnormal behaviors of vehicles based on video analysis technology. First, it uses ViBe (Visual Background Extractor) method to get the foreground image. It tracks the corners by using the Lucas-Kanada optical flow method, getting the corners velocity and direction information. Then, it uses the mean shift method to cluster the two motion features to get the statistical histogram after clustering. Finally, it judges the abnormal behavior of the vehicle with the Euclidean distance of the motion characteristic entropy and the two motion characteristic scalars to the cluster center. The experimental result shows that the two methods can detect vehicle's abnormal behaviors accurately and in real time.

Abstract:In RPKI (Resource Public Key Infrastructure), RP (Relying Party) downloads and verifies certificates and signed objects (ROA, Manifest, Ghostbusters) from repository, and then processes those valid ROA objects into authorized relations between IP addresses and AS number that is used to guide the BGP routing. In the current implementation, the certificate verification module recursively finds the parent certificate of the certificate to be verified through the database query to construct the complete certificate chain and complete the final verification by OpenSSL. Because of the large number of certificates in the RPKI system, the method based on database query is inefficient. Combining the characteristic of RPKI running mechanism that transfers the calculation cost from the BGP router (user) to the RP server (server) and the idea of “space-time tradeoff”, we can read information of certificates into memory to reduce the time consumption of I/O. Based on the ideas above, combined with the characteristics of the time complexity that finding item in hash table is optimal O(1), we design and implement an optimization method of RPKI certificate validation based on hash table. The experimental results show that the average time acceleration ratio is 99.03%, 98.45%, and 97.48% in the three designed scenarios, which has effectively reduced the time consumption.

Abstract:Display of power grid equipment on map is the basic function of power GIS application. According to the statistics, in most of our provinces, the amount of power grid resource data has reached tens of millions or even hundreds of millions. Therefore, the high performance display of massive power grid resource data on web has become a key problem to be solved in the construction of power GIS system. Aiming at the characteristics of GIS data and application of power grid, this paper presents a dynamic loading and rendering method for GIS data of massive power grid resources for Web. This method controls the amount of loaded data by data display rules and power line data compression, thereby reducing the network transmission and drawing pressure. It loads data according to the viewport bound, establishing request cache and data cache on the client to improve the efficiency of data rendering, and uses HTML5 technology for multi-threaded data analysis device symbols rending. It is proved that this method can significantly improve the fluency, interactivity and flexibility of web client-side displaying for massive power resource GIS data, and has greatly improved the user experience.

Abstract:In order to effectively improve the activity classification efficiency in body sensor networks, a maximum likelihood sparse representation algorithm based on K-SVD is proposed in this study. Firstly, all of activity pattern training samples are grouped according their classes to be trained, respectively. The mutual interference among different groups in the process of training can be avoided and sub-dictionaries for every class can be obtained. Then, these sub-dictionaries are used to construct an over-complete dictionary. And the dictionary is able to sparsely represent the testing samples precisely. The sparse representation coefficients are precisely approximated by maximum likelihood sparse model and the recognition result of testing samples are determined by the coefficients. The experimental results show that the proposed algorithm is able to obtain the optimal dictionary and the method based on maximum sparse representation can precisely estimate the representation error of testing activity samples. The accuracy of the proposed algorithm is obviously better than some conventional sparse-representation-based activity recognition algorithms. The proposed algorithm is able to effectively improve the activity pattern classification efficiency in body sensor networks.

Abstract:Data mining technology is an effective approach to resolve the problem of abundant data and scanty information. Outlier mining is one of the main research topic in the field of data mining, and it has been widely used in network intrusion detection, line card fraud, spam analysis, gene mutation analysis, etc. In high-dimensional data, the data volume and high dimension affect the effects of outlier data mining and efficiency seriously. In view of the high dimensional data, this study adopts the KNN implementing a distance-based outlier data mining algorithms under the MapReduce programming model by defining the “solving set”. Using artificial data set and UCI data set, the influence of parameters on the algorithm performance is discussed under different conditions in the experiment.

Abstract:According to the X-ray weld image collected by a steel pipe factory and the study on diversity and morphological variability of weld defects, a weld defect identification algorithm based on rotation invariant HOG feature extraction is proposed. First of all, we classify different types of defects detected to extract ROI of each image, all of which constitute the defect samples required by the experiment. By means of scale transformation and circular cell division, we obtain HOG characteristics with scale invariance and rotation invariance. Then all the sample features are reduced by PCA dimensionality reduction. The dimension is determined by the contribution. Finally, the LSSVM model is used to identify the defects. By studying the effect of block overlap on the recognition accuracy rate, it is found that the higher overlap range, the higher correctness in a certain unit. The algorithm improves the accuracy of defect recognition by improving the traditional HOG feature extraction method.

Abstract:Based on the daily, weekly, and monthly market data of Shanghai A shares, this study uses statistical methods to carry on the data mining research, in order to learn the influence of different horizon, time scale, and stock industry on the distribution of stock returns. From single stock section, it performs the test of normality for the density distribution of price yield and analyzes the relationship between its distribution characteristics, and the circulation market value, the industry category of the stock and the time scale (day, week, and month) respectively. From the unit time section, the study analyzes the relevant statistical characteristics of the mean and volatility of the yield of the stock portfolio. The results show that the variance of the mean of the yield of the stock portfolio is much larger than that of the single stock section because the correlation between the stocks is much larger than the correlation between the time. In addition, the volatility of the stock set also has long-term memory characteristics.

Abstract:In view of the low efficiency and lack of real details in the process of the fluid interaction simulation based on SPH (Smoothed Particle Hydrodynamics), an interactive simulation method based on position-based fluid is proposed to simulate rigid body tool and fluid. This method is improved on the basis of the traditional SPH algorithm, and the interactive process is simulated in real time based on CUDA parallel computing platform. Then, the real-time output interaction force is combined with force interaction device. The experimental results show that the interaction force in the simulation process is in accordance with the expectation, and the continuity and stability of the interaction force are verified under the premise of ensuring the accuracy of the fluid simulation.

Abstract:To measure the size of target in three-dimensional scene under a single camera, this study uses a three-dimensional environment distance function and optimizes the optical axis parameter OO'. A method of enlarging the measurement range is proposed, which is extended to measuring on the vertical surface. The experiment is based on the Raspberry Pi 3B platform with Raspberry Pi Camera. The results of experiment show that the optimization of OO' improves the measurement accuracy and the method proposed extends the measurement range. The algorithm is simple yet effective and can achieve the performance more flexibly and steadily.

Abstract:The construction of feature vector is a key issue for protein secondary structure prediction. In the present methods, only the BLOSUM62 matrix is taken into account, which neglects the amino acid mutation of protein in the evolutionary process. In this study, we propose to construct feature vector by combining PSSM matrices of different evolutionary times, which cannot only reflect the position information, but also reflect the interaction of amino acids. Based on the feature vector, logistics, randomforest and M-SVM_CS models are utilized to predict protein secondary structure on the public datasets (RS126, CB513, and 25PDB). The experimental result demonstrates that the method can achieve a better performance than traditional methods.

Abstract:In consideration of the nonlinearity of SO₂ in circulating fluidized bed boiler, a prediction model of SO₂ emissions in flue gas based on support vector machine is proposed. It is complex to directly search the parameters of support vector machine regression, so a method combining single variable search and grid search is applied. The simulation shows that the prediction model of SO₂ emissions in circulating fluidized bed boiler based on support vector machine has good prediction performance.

Abstract:As a result of the Android system's popularity, the number of malware on it is increasing rapidly. In this study, a static detection method based on multi-feature and Stacking algorithm is proposed, which can make up the shortcomings of the two aspects, i.e., based on single feature and single algorithm. Firstly, this study uses a variety of feature information to compose the eigenvector, and uses the ensemble learning algorithm of Stacking to combine Logistic, SVM, k-Nearest Neighbor and CART decision trees. Then, classifiers are generated through training samples. The experimental results show that the recognition accuracy is up to 94.05% compared with the single feature and single algorithm, and the classifier has better recognition performance.

Abstract:A sales forecasting method based on random forest algorithm and Adaboost method is proposed. Firstly, by analyzing the characteristics of the sales factors, the characteristics and dimensions of the training data are determined. Then, the feature data is trained by the random forest algorithm based on Adaboost, and the steps of the prediction algorithm are presented. Finally, the experimental results show that this method can greatly enhance the performance of random forest algorithm, and has a high prediction accuracy, as well as a good performance of generalization.

Abstract:In order to detect the abnormal crowd behavior in video surveillance in real time and more accurate, this study proposes a method of dynamic interframe space updating based on the Pyramid LK optical flow. The algorithm dynamically updates the interframe interval by extracting the crowd motion information, and then detects the crowd motion information at the interframe interval. In this way, the algorithm does not only preserve the advantages of the traditional algorithm in detecting crowd motion information, but also improves the efficiency. Finally, the algorithm identifies the abnormal crowd behavior by acquiring the intersection density and energy information of the crowd motion vector. By testing multiple videos, the test results show that the algorithm can identify the abnormal crowd behavior in the video with high accuracy, and also effectively improves the running speed.

Abstract:In the cloud computing environment, the execution environment of network security protocols becomes more complex than ever before. The use of Web security issues open license agreement can improve the security of information sharing. CPN (Colored Petri Net) is employed to model OAuth protocol. The analysis of authorization code pattern in OAuth protocol adopts simulation tool named CPNTools. The experimental results show that the authorization code pattern can be verified and authorized based on token. Authorization token injection attack can be prevented in this way.

Abstract:In order to study the correlation between infrared and visible image fusion quality evaluation indexes, the similarity of the fusion quality evaluation indexes of infrared and visible images based on cluster analysis is put forward. In this paper, eleven evaluation indexes are listed, and correlation matrix are respectively established by using Spearman rank correlation coefficient and grey correlation degree. Through the analysis of change rate of threshold, the optimal clustering threshold is selected and the evaluation index of the degree of similarity is given. In the experiment, 10 groups of commonly used infrared and visible fusion images are respectively evaluated by 11 evaluation indicators, and then 11 evaluation indicators are divided into 6 categories by cluster analysis method. The results can be used as the basis for selecting a reasonable objective evaluation index set.

Abstract:Cloth simulation is aimed to generate realistic cloths by the computer. It has a wide application prospect in the visual reality, fashion animation and other fields. In order to avoid the excessive deformation and improve the accuracy of the cloth, we present a technique for strain limiting that dynamically re-calculates anisotropic strain limits. We discretize the cloth model as a triangular mesh, and define the inner-product inverse projection for computing the limits for each principal axis of deformation after remeshing in each frame. It ensures that the local limits of every triangle are adaptive to the global limits. And we propose the improved method to solve the optimization problems by strain limiting. The experimental results show that our technique can effectively improve the accuracy of strain limiting and enhance the realism of the cloth.

Abstract:Text sentiment classification is a popular subject of natural language processing and the crucial problem in product evaluation. Based on semantic relationship of word vector and sentence vector and the impact of user information, product information to text sentiment classification, Cosine Similarity Long-Short Term (CSLSTM) network is proposed. CSLSTM considers attention mechanisms of user information and product information in various semantic levels. And it involves a effective initialization method in hidden level weights of word-level attention matrix according to similarity of word vector and sentence vector. The competitive results are derived from three sentiment classification datasets, Yelp13, Yelp 14, and IMDB.

Abstract:Feature extraction is of great significance in workpiece recognition. In this paper, the image preprocessing is performed on the original image obtained by gray scale transformation and smooth denoising. An improved method of feature extraction is proposed. The SURF algorithm is an accelerated version of the SIFT algorithm, which cannot only ensure the stability of the detected feature points, but can also to a large extent speed up the extraction of the characteristics of the time. It can meet the real-time needs of the workpiece recognition process. The feature recognition method based on the improved SURF algorithm is used to identify the workpiece. Experiments show that the improved feature matching method is accurate for workpiece identification and the speed is good.

Abstract:With the high pace of internet technology, microblog, an opening free social network, has an awful lot of active users. However, the number of sina microblog users is very large and the personal information is not always true, leading to the situation that it is hard to label the user's gender. In this study, multi-view and tri-training learning method are used to solve these problems. First three different views are constructed and three different classifiers are trained with a small number of labeled samples. And then three different classifiers are trained repeatedly by unlabeled samples. Finally, we integrate three classifiers into one to judge the user gender. We use the real user data and find that the classifier using the multi-view and tri-training learning is better than the performance of the single view classifier and needs less labeled data.

Abstract:Weld defect detection is a key link to ensure the quality of welding. The problem of X-ray of weld defect detection has been widely studied with the rapid development of industry and urgent demand. However, because of imaging methods, influence of casting material and other objective factors, X-ray image noise background, low contrast, brightness uneven and weld edge blur, which make use of computer to weld defects automatic detection accuracy is not very ideal. Aiming at these problems, a new method is proposed to detect weld defects in this paper. Firstly, fast independent component analysis (ICA) is used to reconstruct the X-ray image background with defect. Then, the image is subtracted from its reconstructed image to obtain the difference image，and the method of threshold segmentation is used to extract the defects. Finally, further processing on the extracted results effectively reduces the false detection rate. Compared with other traditional detection algorithms, the proposed method is not sensitive to defect types, so it has better adaptability and versatility.

Abstract:Considering the public sentiment is not comprehensively measured in the existing stock market prediction study, the study proposes a stock market prediction model using social sentiment analysis. First of all, a securities sentiment quantitative method based on heterogeneous graph model is applied for sentiment analysis on social media data, and thus quantified sentiment time sequence is obtained. Secondly, a prediction model based on self-organizing neural network is proposed for the stock index prediction by using sentiment sequence and the quotation index sequence. The experimental results on the domestic stock market and social media data sets show that the proposed model has improved by 15% and 12% over the BP (Back Propagation) neural network in the prediction error and accuracy respectively, which can better predict the stock market.

Abstract:Interfered by a variety of factors, indoor positioning has been a research hotspot in wireless network. To improve the indoor positioning effect, aiming at the problem that the neural network has in indoor positioning accuracy of the wireless network, this paper designs a wireless network based on artificial neural networks. The first indoor wireless network collects relevant information, extracts indoor positioning data, and then uses neural network for data learning. It sets up a wireless network positioning model to improve the defects of the neural network. Finally, the simulation is carried out on the Matlab platform. The results show that the improved neural network overcomes the limitations of the traditional indoor localization methods, and achieves higher indoor localization accuracy of wireless networks. Moreover, the indoor localization efficiency has also been improved significantly.

Abstract:The dorsal line of cattle is one of the most important indexes in the identification and classification of Xinjiang brown cattle, and the level of the back line reflects the growth of cattle, which is an important index for selection and breeding. Based on the identification standard of linear type of Xinjiang brown cattle, taking cattle side image (mainly from the chest to the cross step image) as the research object, this paper uses digital image processing method first to process the image binarization, and then to detect the edge of image binarization, realizing the automatic extraction of the edge points of the back line cattle. Finally, through the analysis of the data of the edge of the back line, it gets the automatic classification of the cattle back line, and divides the line into five grades: 45, 35, 25, 15, 5. The higher the score is, the better the growth of cattle is. The experiment shows that the algorithm is effective and feasible, and can get the result of automatic classification of Xinjiang brown cattle back line fast and precisely.