This PDF file contains the front matter associated with SPIE Proceedings Volume 12510, including the Title Page, Copyright information, Table of Contents, and Conference Committee list.

This paper makes bibliometric analysis for related indexes in the field of rare earth-magnesium alloys, for example, output scale of articles, academic influence, discipline layout, and international cooperation, and compares China’s strength with other countries around the world to obtain development trend of the technology in the world, thus formulating related development strategy and discipline policy for the technology and providing useful reference for corporate investment decision-making and researchers in selecting research direction.

Based on the enterprise sample data and the national database, this paper analyses the carbon dioxide emissions in Anhui Province, including the carbon emissions in key areas, the classification of peak situation and the trend of carbon reduction by area. The results show that the power industry, cement industry and steel industry in Anhui Province have the largest proportion of carbon emissions. According to the characteristics and development trend of the industry, the carbon emissions of the power industry have shown a continuous upward trend, and the carbon emissions of the cement industry and the steel industry have gradually entered the platform period. In the future, the increment of renewable energy will become an important support for power supply and carbon emission reduction in Anhui Province in the long term, including clean power from outside the region.

The development of the new energy automobile industry is an important direction to accelerate the construction of a powerful automobile country, and it is also an effective means to break the constraints of energy and environment, which is of great strategic significance to China. In this article, based on the perspective of the development of the overall new energy vehicle industry, we analyze the market size and development stage of China's new energy vehicle industry, and explore the characteristics of the industry's development. Further, this paper analyzes the industry risks of the future development of new energy vehicles under the background of the imminent withdrawal of government financial subsidies and the transition of the new energy vehicle industry from policy-driven to market-driven and provides theoretical support for the healthy and sustainable development of Chinese new energy vehicle industry.

Company decisions cannot be implemented through the subjective will of leaders; therefore, market research is required to help determine the approach. By continuously comparing and combining information, industry research will carry out its purpose. In this work, we first use Python to do the data pre-processing, then contrast the most pertinent variables. Regression analysis, comparative analysis, and other techniques are being used to analyze the responses. In this paper, using wine sales as an opportunity to highlight overall spending. The findings indicate that people's purchasing preferences are greatly influenced by their quality of qualification and revenue. Discovering if there is a connection between the factors is also another goal.

This article looks at the impact on biological and medical companies in the context of covid-19. Through linear analysis, data model plots, and variance comparisons, the study finds that the economy of the traditional medical sector continues to decline and eventually plateaus, however, the pharmaceutical as well as biological research sectors, which are associated with the epidemic, continue to rise. The advantages and disadvantages of combining the internet and healthcare are presented. Although internet healthcare is a more convenient way for people to see a doctor. Some people can even see a doctor for treatment without leaving home, but diagnosis lacks accuracy and large testing machines. And a more rational approach to healthcare must still be researched in the future.

Real estate investment risk is an unpredictable factor affecting development profit, so it is very important to analyze real estate investment risk. In this paper, by considering the real estate decision-making stage, preparation stage, construction stage and the stage of management risk, real estate investment risk analysis index system is established, based on the practical engineering project construction sample, risk factors and expectations index based on support vector machine is the response of the relationship, and the actual project risk analysis.

In recent years, driven by the huge profits in the illegal cultural relics circulation market, smuggling and excavation of cultural relics have been repeated, and the situation of heritage crimes has become more and more serious. It is important to understand the occurrence pattern of excavation-type heritage crimes and construct a time-series prediction model of excavation-type heritage crimes to prevent them. This paper uses the random forests algorithm to construct a time-series prediction model of heritage crimes, which effectively solves the problem of poor timeliness of traditional prevention methods and is an attempt in the field of heritage crimes prediction. This paper constructs a time-series data of heritage crimes at several time scales and finds that the model has the best prediction effect when the time step is set to 30. It suggests that there may be a certain pattern of occurrence of excavation-type heritage crimes at the monthly scale. The findings of this paper are expected to provide decision support for the deployment of prevention and control resources for protected heritage units.

Oligopoly is a prevalent market structure to be seen in the market economy around the world. There is a countable plural of firms in the market, which makes analyzing this market structure more complex than other market structures since firms do not act as a whole firm or a united group of firms. Game Theory is often utilized to analyze their actions; however, current analysis does not include firms’ innovation and government intervention impacts. In our work, we first look at quantitative analysis regarding prices only in oligopoly markets with homogeneous and differentiated goods. Later we analyze firms’ actions qualitatively in a relatively real-world setting by considering innovation and government intervention.

Considering the influences of the COVID-19 disease, systemic risks with respect to the tourism industry and the erratic preferences of the tourists have fiercely affected the performance of machine learning models for tourist trajectory prediction. This paper introduces a noise-reduced and Bayesian optimized light gradient boosting machine (LightGBM) to forecast the likelihood of visitors entering the consequent scenic attraction, accommodating to the variability of tourism attributes. The empirical evidence of tourism data in Luoyang City Hall from March 2020 to November 2021 illustrates that our practice surpasses the baseline LightGBM mechanism as well as a random search-based technique regarding prediction loss by 5.39% and 4.42% correspondingly. The proposed research demonstrates a promising stride in the improvement of intelligent tourism in the experimental area by enhancing tourist experiences and allocating tourism resources efficiently, which can also be smoothly applied to other scenic spots.

In the context of the dual-cycle paradigm, the paper constructed a comprehensive evaluation system for high-quality foreign trade development in China from three dimensions which are high-quality foreign trade capacity, high-quality foreign trade environment, and high-quality foreign trade effects. By combining the entropy method and the TOPSIS method, we comprehensively evaluated the high-quality development level of foreign trade in 251 prefecture-level cities in China and analyzed the specific performance of foreign trade in the past 20 years from macro and micro perspectives. The result reveals that Chinese foreign trade development was progressive, and gradually developed in the direction of high quality on a relatively stable fundamental level. But compared with double follow bad policy guidance, China's foreign trade dependence is still at a higher level. Therefore, its development trend is worth investigating. The foreign trade level of various coastal provinces and cities with developed foreign trade was far ahead of other provinces and cities. From the micro level, the development of foreign trade in cities of various sizes has been rising steadily in recent years. However, it can be found that there is still large room for the improvement of the high-quality development of foreign trade in some small and congested cities. It can be seen that the development status of Chinese foreign trade regions was uncoordinated and unbalanced. Finally, based on the above studies, this paper presented the optimization ideas to promote the high-quality development of China's foreign trade under the guidance of the double-circulation new development pattern.

Analyzing the research changes of interrogation methods in our country and revealing the hot topics and research trends in this field is of great significance for improving interrogation strategies and realizing the transformation of interrogation methods. This thesis uses the papers on the subject of "interrogation method" in CNKI database from 2002 to 2021 as the data source and uses CiteSpace software for visual analysis. The results show that the hot topics in the research field of interrogation methods include investigator, evidence, human rights protection, interrogation skill, etc. While the research trends mainly focus on major case, repeated confession, confession psychology and so forth. As the current related research is still weak in terms of academic cooperation, research methods, and foreign reference, it is recommended to increase academic exchanges and cooperation in the future, strengthen interdisciplinary research, and learn from international advanced experience based on our country's national conditions, so as to expand the breadth and depth of research on interrogation methods.

GFSR is a common parallel Random Number Generator (RNG). In this paper, the high-performance parallelization of GFSRc(521,32) on Field Programmable Gate Array (FPGA) is realized by using the hybrid method. The serial and the parallel random number sequence of different numbers are tested using the Diehard test, graphical test and application test. It is found that the random number sequences parallelized by GFSR (521,32) have good randomness. For 100 million random numbers generated, speedups of 5755× over single-threaded CPU platform and 453× over 32-threaded MIC platform are achieved. Experimental results show that GFSR (521,32) is a good random number generator suitable for large scale parallel Monte Carlo simulation on FPGA.

A deep shale gas reservoir has become a research hotspot in the field of unconventional oil and gas resources in recent years because of its huge reserves and development potential. In view of the characteristics of low permeability, heterogeneity and anisotropy of deep shale gas reservoirs, this paper adopts the Markov Chain-Monte Carlo (MCMC) algorithm based on statistical theory to reconstruct 3D digital cores of shale samples from deep shale gas reservoirs. Compared with traditional methods, such as statistical function, for testing the validity of MCMC algorithm, this paper verifies the validity of MCMC algorithm by calculating gas-water relative permeability curves of the reconstructed digital cores, where MCMC algorithm is applied to obtain different digital cores with similar gas–water relative permeability curves based on the same 2D scanning electron microscope image.

In this paper, the regional economic system is regarded as a composite system composed of port logistics and economic subsystems. The synergy theory is used to build the synergy development model of regional port logistics-economic system to analyze and calculate the synergy development degree of regional port logistics and economy. Taking Chongqing port logistics and economic system as an example, the statistical data from 2000-2020 were collected to validate the model. The study shows that the scientific calculation of the synergy degree can be completed by the model to reveal the synergy development level of regional port logistics and economic development, and Chongqing city has a strengthened synergy development relationship between port logistics and economy year by year, which has now reached a highly synergy development state.

In this paper, through the analysis of the mid infrared spectrum data of Chinese medicinal materials, we establish a mathematical model to classify the given Chinese medicinal materials, analyze the characteristics and differences of the medicinal materials from different places, and identify the origin of the medicinal materials. Based on the near-infrared and mid infrared data, we identify the origin of a kind of medicinal material. Moreover, according to the near infrared spectrum data of several medicinal materials, we identify the classification and origin of medicinal materials.

Shuikou Township, Huzhou City, Zhejiang Province is selected as the research area, and the rural non-use value of Shuikou Township, Huzhou City in 2022 is quantitatively studied by using the open double boundary dichotomy Conditional Value Method (CVM) and the Willingness To Pay (WTP) survey. The research results show that: 1) in 2022, the average willingness to pay value of the rural nature of the tourist destination is 82.74 yuan / person without considering other explanatory variables. After taking into account some explanatory variables such as personal social and economic attributes, the average willingness to pay value is 90.91 yuan / person, of which existential value and choice value are the main forms; 2) the potential value of the rural nature is huge, and the protection of the rural nature is very important to the rural tourism of Shuikou Township, which should attract the attention of rural tourism operators.

The barefoot footprint is a common trace material evidence in the crime scene, and it is an important basis for investigating and solving the case. However, the identification of barefoot footprint is mostly realized by manual operation, which is easily affected by subjective factors. It is difficult to compare and identify a large number of barefoot footprint samples. This paper puts forward a variety of barefoot similarity measurement methods, and uses MATLAB software to realize the automatic segmentation, extraction and comparison of the physical and morphological features of barefoot footprints. After image binarization, open operation and closed operation by corrosion and expansion processing, convex hull operation, edge detection and so on, by using many kinds of similarity measurement methods, the features of the footprint can be extracted completely and calculated accurately. According to the measurement principle of different "distance" in computer vision image processing and machine learning, an algorithm is designed to measure the similarity of 120 barefoot footprint samples imprinted by the same person and different people. The accuracy of the algorithm is tested by comparing the calculation results, so that the algorithm can be applied in practice.

From a perspective of research output, we use the papers in the human resource management field in the Web of Science database as the data basis and proposes an analysis method of the development trend of human resource management based on the keyword co-occurrence network. The critical research content in this field is obtained from the primary geometric feature and network centrality of the keyword co-occurrence network. By extracting network subgraphs, the paper analyzes the research hotspots of three crucial countries, including the United States, the United Kingdom and China. Finally, based on different time windows, the research trend in human resource management is drawn.

Based on the traditional comprehensive measurement system of economic vitality, this study adds two dimensions of enterprise vitality and talent science and technology, and uses the data of 16 districts in Beijing to analyze the economic development of these districts. In order to more accurately verify the reliability of the evaluation index system, firstly, the analytic hierarchy process, grey correlation degree -TOPSIS method and coefficient of variation method are used to calculate the economic vitality value and ranking of all districts. The results obtained by the three methods are highly consistent, which proves the effectiveness of the index system. Finally, using the initial ranking, the comprehensive ranking is obtained by the list aggregation ranking method. The empirical results show that the economic development of Beijing is not balanced. The economic vitality of the district located in the center of Beijing is high, while that of the surrounding suburbs is low. Taking the first ranked Haidian District as an example, this paper selects the indicators with high weight in the evaluation system and makes an in-depth analysis of Haidian District's own advantages and policy orientation. In order to enhance the overall economic vitality of Beijing, the government should increase investment in scientific and technological innovation, strengthen policy support, attract more talents and enterprises, and create new vitality for the urban economy.

According to the stock data of photovoltaic building integration related companies of China Southern Power, a five-day, ten-day, and 20-day moving average model is established for the photovoltaic building integration sector index, and the model is drawn according to the established model on April 1, 2019. The moving average of the sector index through April 30, 2021. Using the data from May 6 to May 28, 2021, the error analysis of the established model is carried out and the model is revised. According to the revised model, the future development trend of the sector is predicted, and the 20 trading days after May 28 are given. daily moving average, 3-week weekly moving average, and 2-month monthly moving average. A technical indicator used to observe the trend of changes in security prices, helping traders to confirm the existing trend, judge the trend that will appear, and find the trend that is overextended and is about to reverse.

Quant investing is very popular nowadays. To predict returns, there are lots of algorithm and models for financial market trading. In this case, with some new methods, quantitative researchers’ ability to forecast an investment’s return. In this paper, a new dataset obtained from a Kaggle competition called Ubiquant Market Prediction will be introduced and some classical method and model will be tested on it. This dataset has following characteristics. Firstly, its’ target is return rate instead of the return, which means it could adapt to different scales of investment. Secondly, 300 dimensional anonymous features extracted are used in prediction so that this method could be applied in the prediction of different kinds of investments which have different kinds of features. Thirdly, this dataset considers the types and time periods of investment behavior. For the reasons above, the dataset is quite valuable. The classical methods tested in the experiment include Support Vector Regression (SVR), ensemble learning, k-Nearest Neighbor (KNN), Deep Neural Network (DNN), random forest, Stochastic Gradient Descent (SGD) regressor, passive aggressive regressor and Multilayer Perceptron (MLP) regressor. These methods are tested on this dataset and their performance are compared in the result and discussion section, aiming to figure out whether these classical approaches could have a good result in the dataset.

Stock prediction is a vital and difficult topic where researchers have explored plentifully. Because of some characteristics of the stock market, like high noise and strong nonlinearity, it is quite hard to forecast the price of stock accurately. Long Short-Term Memory neural network (LSTM) uses input gate, output gate and forgetting gate to control information, and shows excellent performance in processing temporal information.

With the increasing exploitation of marine resources, there is worldwide concern about the marine environment. It is a hot topic, how to assert the marine pollution. By concentrating on dissolved oxygen and suspended solids in offshore water of yellow river mouths, this paper establishes a model of marine pollution assessment by a compositional interval-valued fuzzy relation with Bandler-Kohout subproduct. And a real-life example is shown for the sake of illustrating the efficiency of our proposed model.

With the development of science, the technology of the internet has also presented new trends. With the integration of the internet and industry, the era of "Internet+" has also arrived. During this period, the development of cross-border e-commerce in China has ushered in new opportunities. However, as a new form of international trade, there are also many potential risks. The global outbreak of the new crown epidemic in early 2020 further accelerated the pace of cross-border e-commerce to replace traditional trade. In recent years, cross-border e-commerce has become the new way of selling trade. China's cross-border e-commerce has also made leaps and bounds, with an expanding range of exports and an increasing role in promoting the domestic economy. This essay is based on the literature about cross-border e-commerce under the "Internet+" from 2015 to 2022 in the CNKI database and uses CiteSpace technology to visualize the current development of cross-border e-commerce under the "Internet+" through a visualization method. The study shows that China's cross-border e-commerce development has been greatly improved. Meanwhile, the scope of research in the field of cross-border e-commerce in China has been expanding and becoming more mature and stable, and most researchers have made outstanding contributions to the field. The research hot spots in this essay are highly related to the background of the times. Through the visualization and systematic study of CiteSpace in the field of cross-border e-commerce, this essay shows the information on postings, keywords, and hot spots, which provides certain references for the development of cross-border e-commerce in China.

Regression analysis has always been a popular method of data analysis because of its good interpretation of causality. Variable selection can effectively reduce the complexity of the model and improve the modeling accuracy. As a classical variable selection method, Lasso can realize variable selection and parameter estimation at the same time, but the optimal regularized parameters may lead to inconsistent variable selection results. Adaptive Lasso improves the consistency of variable selection by adopting the least square estimation of coefficients as the weight correction regulars. However, the design matrix of least square estimation is not invertible for ill-conditioned data. In this paper, a modified Adaptive Lasso based on regularization weights and its ADMM regularization algorithm are proposed, taking the Ridge estimation of coefficients as the weight parameters to modify the Lasso regularization terms. Numerical experiments with sparse coefficients, large coefficients, grouping effects, and polynomial regression simulation experiments are carried out to compare the proposed method with Lasso, Adaptive Lasso, Elastic net and the Adaptive Elastic net. The results show that the proposed Adaptive Lasso based on regularized weight has higher prediction accuracy and more sparse variable selection, which effectively improves the accuracy of regression model.

Due the redundancy of a large number of user data, how to use data mining to efficiently and scientifically establish the loan default model is an important issue for reducing bank loan loss. In this paper, logistic regression is adopted to identify the nine significant influencing factors of loan defaults. Further, differences among multiple groups of different samples are studied with the methods contrast analysis and Analysis of Variance (ANOVA). Through factor analysis, eight common factors are extracted, which could explain 80% all of the original variables. This study can provide some implications for banks to use when reviewing loan applications.

The digital transformation of power grid enterprises is a requirement and concrete implementation measure of China's "14th Five-Year Plan" for digital economy, which provides continuous impetus for the flourishing and healthy development of China's digital economy. Power grid enterprises use data elements to provide data services to different targets and obtain revenue. How to reasonably allocate the revenue is an important factor to motivate them to carry out data services and digital transformation. In this paper, we study the pricing strategy of data services of power grid enterprises under different scenarios and develop a revenue allocation mechanism according to different data subjects. This study can provide suggestions for the revenue allocation practice of data asset services in power grid enterprises and help them better implement the digital transformation strategy.

In the background of the era of big data, patents carry a large amount of important information, and the analysis and mining of patent data information is conducive to technological progress and development. With the increasing emphasis on environmental protection in the world, electric vehicles have developed unprecedentedly. In this paper using the 1976-2020 Derwent patent database of electric vehicle patent data, combined with the relevant theory and method of patent measurement, respectively from the industry development trend, patent high yield institution distribution, patent regional distribution, hot technology distribution and the main patent holder cooperation network perspective, analyzes the general situation of global electric vehicle industry development. To provide a reference for promoting the technological innovation and further development of the electric vehicle industry.

The wind turbine operating condition monitoring method has the problem of large errors, therefore, a data mining based wind turbine operating condition monitoring method is designed. According to the spatial position of the wind turbine shaft, the vibration signal of the wind turbine is collected, the input power of the wind turbine and the wind energy utilization coefficient are calculated, the operating condition identification model is constructed, the training data set and the test data set are obtained by using sampling, and the monitoring model is optimized based on data mining. Test results: The mean values of RMSE, MAPE and MAE for the wind turbine operating condition monitoring method in the paper are less than those of the other two wind turbine operating condition monitoring methods: 0.1289~1.3159, respectively, indicating that the overall error of the wind turbine operating condition monitoring method in the paper is smaller after making full use of data mining techniques.

In order to ensure the structural health of civil aircraft in service, health monitoring systems that can collect large amounts of data to reflect the structural condition of civil aircraft have been put into use, which has led to the "big data" era of structural health monitoring of civil aircraft. The data collected by health monitoring systems is characterized by large capacity, diversity and real-time. The research of advanced theoretical methods to extract effective information from health monitoring data and efficiently and accurately identify the structural health status has become a new problem in the field of structural health monitoring of civil aircraft. Theories such as data-driven deep learning and machine learning, as the latest research results in pattern recognition, have achieved fruitful results in various fields of big data processing with powerful modelling and characterization capabilities. Combining the characteristics of health monitoring data with the advantages of data-driven theoretical approaches, this paper proposes a new method for processing health monitoring data of civil aircraft structures. The method uses machine learning and deep learning algorithms to train structural state prediction models. The advantage of this method is that it is able to extract and make intelligent judgements on the state of the airframe without relying on the processing technology and engineering diagnosis experience of a large amount of collected data. This paper also compares the accuracy of various machine learning and deep learning algorithms and explores the differences in the weight of the acquisition signals and the dimensionality reduction of the acquisition data. Experimental results show that the method achieves accurate identification and prediction of the number of structural cycle take-offs and landings and crack expansion states in segment-level fatigue testing of aircraft.

An accurate passenger flow prediction is important for subway station operators and passengers, because it can reduce the congestion of subway stations, ensure passengers’ safety, and reduce passengers’ waiting time. To get an accurate prediction result, an appropriate time granularity selected in the prediction is necessary. The primary objective of this research is to find a time granularity as short as possible, and this time granularity can also ensure the stability and regularity of passenger flow. This paper analyzes the 25-day passenger flow data of 81 subway stations in Hangzhou, takes Pearson correlation coefficient as the index of the regularity, and analyzes the regularity of passenger flow in time granularity of five minutes, ten minutes, 15 minutes, 20 minutes, 25minutes, and 30 minutes under three different passenger flow modes on Saturday, Sunday and working day. The result shows that under the three different passenger flow modes, the passenger flow has strong regularity at all selected time granularities. Specifically, when the time granularity increases from five minutes to 15 minutes, the regularity increases significantly, and when the time granularity is greater than 15 minutes, the growth of regularity has significant slowdown.

In recent years, privacy data has been stolen frequently, and the security of privacy data has been concerned by society. In the big data environment, privacy is facing unprecedented challenges, and some traditional privacy protection technologies are facing failure, so how to choose a reasonable privacy protection technology is a challenging task. To solve the problem of a large amount of data theft in the application of traditional methods in the privacy data security protection, this paper proposes research on privacy data security based on multi-party computation. The hash function is used to encrypt the private data, and the data is stored in the block chain in the form of a private data encryption file. Based on the theory of multi-party computation, a secure multi-party technology protocol is designed, and the protocol is used to verify the identity of the participants in the private data transmission to realize the security protection of private data. The experimental results show that the amount of privacy data stolen by the application design method is less than that of the traditional method, which has important application value for the security protection of privacy data.

In information theory, information entropy is used to measure the amount of information, i.e., the amount uncertainty of random variables. And it has been adopted in psychological experiments and behavior analysis. In this paper, based on the survey data from Chinese thousand village survey and using the technique of information entropy, this paper defines the measurement of village’s ecological environment and shows that raising the environmental consciousness of villagers and village committees can significantly improve the ecological environment of the area because people will transform their environmental awareness into action. Also, the paper finds that the village committee’s action on ecological restoration has a significantly positive effect on the villager’s environmental awareness and willingness to protect the environment. But the influence of people’s environmental consciousness is limited to domestic pollution and has little effect on enterprise or agricultural pollution which are closely related to economic profit.

With the progress and development of the power system, the automation, information, and intelligent technology of the power system have received more powerful technical support. However, as an important "network battlefield" in many countries, the security problem of the power monitoring system is becoming more and more prominent. Aiming at the problem that the threat situation of power databases is obviously rising under the action of malicious attacks, this paper proposes a security protection method for power system database based on the cloud platform. A public key and a private key are set by using an RSA encryption algorithm to encrypt the database; based on the Amazon Web Services of the cloud platform, a large prime number is selected from the Simple Storage Service of the cloud platform as a client key in a random selection mode. The database judges the legality of the access according to the relationship between the received verification key and the public key and private key set by the RSA encryption algorithm and makes feedback on the access after judging that the access is legal. The test results show that after the implementation of the design method, the threat situation value of the database is always less than 0.50 under different attacks, which can play a role in security protection.

Encountering with new social environment and data technology, the Police Commanding & Allocating System need grasp the opportunity to utilize infinite cyber data resource and integrate data technology with modern police activities. Urgent command delivered by Control Centre usually makes policemen on duty feel overwhelmed and exhausted. Correct prediction of 110 Amount Alert can help policemen on duty form reasonable psychological expectation and make enough preparation in advance. As is known, weather of a day in the near future is easily and correctly predicted. If correlation of 110 Alert Amount and weather factors is found, we can roughly predict 110 Alert Amount in the future one or two days based on weather forecast.

Categorization is an important and common practice in medical research, and finding cut-off points is a direct way of classification. The existing methods are mainly about computing cut-off points for only one variable. This paper proposes an algorithm to select cut-off points for multiple variables in binary outcome settings. This algorithm extends the search for cut-off points of one variable to multiple variables and is applied to the Breast Cancer Coimbra dataset. It shows a good classification effect and gives the confidence intervals of the cut-off points of the prognostic variables.

Random Forests (RF) are a popular machine learning method for developing variables selection models. However, it may suffer from a lack of better model interpretability compared with traditional models, such as Logistic Regression (LR). In this paper, we propose a predictors selection strategy based on Stepwise Random Forests and Logistic Regression model (SRFLR) and validate it using HCV data provided by UCI machine learning repository platform. The Synthetic Minority Oversampling Technique (SMOTE) algorithm is adopted to deal with the problem of imbalance class in the dataset. The results demonstrate that the proposed SRFLR can obtain more predictors while preserving better prediction ability than LR alone, which will offer some references for clinical researchers to select relevant disease predictors.

The operation of the network is often accompanied by network attacks, and as the network begins to gradually enter the 5G era, the intensity of network attacks and attacks are also rising significantly, of which Distributed Denial of Service (DDoS), as the mainstream attack of network attacks, has a very high attack intensity and attack frequency, and the traditional detection system is unable to detect the complex DDoS attacks. However, when the detection scheme is combined with artificial intelligence, it becomes efficient and reliable. Over the years, the detection of DDoS attacks has achieved very good results. However, in the face of the advent of the 5G era, higher requirements are put forward for detecting DDoS attacks. Given the good potential and performance of Convolutional Neural Networks (CNN) in classifying images, this work proposes a method to convert network traffic into images and apply the ResNet model for training on the converted data. The work sacrifice some of the accuracy on binary classification, which drops to 99.6% but improve 2% on multivariate classification over previous work to 89%.

The growth of big data usage has led to an increase in the number of big data companies, and the demand for big data talents is increasing. However, because the existing number of big data talents cannot meet the needs of enterprises, it inhibits the development of big data enterprises to a certain extent. The present study used Python web crawling technology to capture more than 70,000 pieces of information about big data talent recruitment from 51job.com. Our team then cleaned, analyzed and visualized the data. Through data analysis, we found that the company did have a huge demand for big data talents, and we assessed the influence of geography, education and salary on the recruitment of big data talents. The skill requirements of big data talents are also recorded. Finally, we offered suggestions and plans as a reference for the training of big data talents in China from the three levels of the government, enterprises and universities.

To solve the problem of financial fraud of listed companies is the fundamental way to maintain the smooth operation of the whole stock market. In order to explore the related problems of the governance of the financial fraud of the listed companies, the influence of the future theory on the strategies of the related subjects is introduced and the "listed company - accounting firm - Prison" is constructed. The static game model between the management departments and the static expansion game model are used to analyze the stability conditions of the main strategy selection related to the whole financial fraud process of the listed companies. The study shows that the absolute value of perceived income of the related strategies is the key to the choice of the strategy. Therefore, it is a common means to promote the establishment of the relevant stable strategy conditions. At the same time, it is more objective and reasonable to eliminate the possible influence of the relevant main strategies on the model, that is, the selection of the reference points, the objective probability instead of the weight function and the introduction of the value function to the related variables.

Along with the gradual development of the international oil market in the direction of financialization, more and more investors are investing and trading oil as a financial asset, resulting in crude oil price fluctuations affecting the transfer of input funds between stock markets. In turn, the transfer of funds in stock markets will trigger changes in a country's exchange rate. Therefore, crude oil is an essential strategic resource, and its price fluctuations are bound to move the country's stock and currency markets, thus bringing systemic risk. Therefore, based on the industry perspective, the SETAR-Copula model is used to portray the nonlinear characteristics between the exchange rate, WTI crude oil futures price index and SSE index, and the sample data are divided into different intervals according to the threshold parameters in the model. The Vine Copula function is combined to measure the conditional value-at-risk CoVaR and risk spillover ΔCoVaR under various combinations of intervals. The study shows that the stability of the exchange market and the stock market in China can weaken the impact of the volatility of the international crude oil market when it is in a fast-rising stage.

House prices have always been a major concern in people's daily lives, not only for individuals, but also for the stability of society as a whole. Accurate prediction of house prices can help people to buy houses with a great degree of reference and help the state and society to control the overall consumer prices. With the development of machine learning, we are able to make greater use of machine learning tools to obtain accurate forecasting of house prices in recent years, in order to attempt to predict the house price, this article utilize the machine learning algorithms. The house price problem is a typical regression problem in machine learning. In this paper, we obtain data from Kaggle, look at the data and find that there are many factors affecting house prices. By pre-processing the original data, engineering and standardizing the features, finally, machine learning algorithm XGBoost is selected to establish a prediction model for housing price prediction. We selected the root mean square error (RMSE) ranging from the logarithm of the expected value to the logarithm of the real selling price as the final result of the forecast. The final results we obtained can help one to judge the accuracy of the housing price forecast.

To solve the jamming decision problem of the non-cooperative multifunctional radar in a complex electromagnetic environment, this paper proposes a radar cognitive jamming decision method based on an incomplete information game. First, the characteristics of current multifunctional radars are analyzed, and a radar jamming decision model is established based on an incomplete information game; then, after quantifying the gains of both sides by considering the jamming effect and radar anti-jamming capability, a jamming strategy selection algorithm is designed for unknown operating mode radars, and the magnitude of the jamming effect is used as the criterion for strategy seeking. Finally, the proposed method is scientifically verified utilizing simulation experiments. Compared with traditional radar jamming decision methods, this method proposed in this paper is more suitable for practical operation in terms of model design and gain quantification, and the increase in the number of radar operating modes has no impact on the accuracy of a strategy seeking; the output of jamming strategy is a pure strategy, which is accessible and apprehensible for equipment operators.

Cultural Heritage is a witness to human history, and its preservation has become a worldwide consensus. However, heritage crime is still rampant, and the protection of cultural heritages still needs excellent efforts. This paper analyzes the crime pattern of cultural heritage from the time series perspective and proposes a targeted early warning framework. The trends of the four heritage crime types were summarized on weekly, monthly, annual and decadal time scales. The upstream and downstream heritage crimes were strongly correlated on the daily time scale, with 90% of years correlated on the scale of 0.01. For different provinces and different types of cultural heritage crimes, a comprehensive early warning model framework is established using a combination of Prophet model prediction and qualitative analysis, and the optimal prediction horizon is given. The analysis results of this paper provide a viable path for applying heritage data to conservation work. It helps to prevent and control heritage crimes and provides ideas for the information construction of heritage security.

In the study of the impact of cross-border capital flows, most scholars at home and abroad focus on the method of linear time series mainly based on the vector autoregressive model (VAR), ignoring the volatility of variables in time series. In order to make up for the deficiency, the dynamic conditional correlation-generalized autoregressive conditional heteroskedasticity (DCC-GARCH) model can be used to study the nonlinear time-varying correlation between variables. With the help of Eviews12 software and the DCC-MVGARCH model, this paper studies the impact of securities markets on cross-border capital flows in China from domestic and foreign perspectives in the context of two financial crises and COVID-19. The results indicate that financial crises affect the correlation between the securities markets and cross-border capital flows. China's stock market is positively correlated with short-term capital flows and negatively correlated with long-term capital flows. Its booming bond market promotes short-term capital flows but fails to affect the long-term capital flows, and China's short-term capital flows are increasingly linked to the volatility of foreign stock markets. Therefore, it is necessary to improve the mechanism for better monitoring and analyzing cross-border capital flows, promote further development of financial supervision, and guide market players to face the securities market rationally.

Formation target track sorting is an important problem in multi-target track processing. Unlike the processing of periodic scans and point tracks of active radar, passive positioning data has the problems of irregular positioning period and difficulty in distinguishing multiple targets by target signal attributes. In this paper, we propose a track initiation algorithm for passive positioning system based on density clustering and Hough transform for the characteristics of passive positioning system. The algorithm first uses the density clustering algorithm to identify the cluster centers of targets with large measurement errors, and then uses the Hough transform method to initiate a track.

The era of "dancing with robots" has arrived, and many industries, including auditing, have begun to explore the "artificial intelligence +" model. In this context, this paper first discusses the opportunities brought by artificial intelligence technology to the reform of the audit model, and then adopts the questionnaire method to analyze the problems faced by the reform of the audit model in the current era of artificial intelligence, and finally puts forward relevant suggestions on how to promote the change of the audit model led by artificial intelligence. This paper has certain guiding significance for promoting the innovative development of auditing models and ensuring that auditing can play a better role in the national supervision system.

Aiming at the shortcomings of Dingo Optimization Algorithm (DOA), such as low precision, less diversity of population and weak global searching ability, an improved dingo optimization algorithm based on periodic convergence factor strategy is proposed. The convergence factor is incorporated into the location update of group attack, persecution attack and sweep of the basic DOA, and the convergence speed and accuracy of the algorithm are obviously improved. By solving the low-dimensional and Galway test functions, the experimental results verify the optimization performance of the improved algorithm, which can be concluded that IDOA has stronger optimization ability and faster convergence speed.

Most of the current trajectory tracking algorithms, which only rely on a single data source sensor for target trajectory tracking, are prone to problems such as inaccurate acquisition of target velocity information and unstable tracking. In this paper, based on the existing 2D and 3D trackers, a trajectory information fusion algorithm based on adaptive weight coefficients is proposed based on the consideration of the distance decay characteristics of point cloud quality (quantity, density, etc.), which effectively improves the accuracy and robustness of the trajectory tracking method, expands the trajectory tracking range, alleviates the target loss phenomenon, realizes the mutual fusion of multi-dimensional trajectory information under the same time and space, and makes up for the shortcomings of a single sensor when tracking the target trajectory.

Aiming at the problems that sparrow search algorithm is easy to fall into local optimum, slow convergence speed and insufficient convergence accuracy, an Improved Sparrow Search Algorithm (ISSA) with weighted mutation and neighborhood perturbation is proposed. Improve the follower position update formula and add the weighted mutation differential evolution algorithm to maintain the population diversity and balance the global exploration and local development ability of the algorithm. Combining attack with reverse search and variable spiral position update, the algorithm perturbs the individual position around the optimal neighborhood, jumps out of the local optimization, and improves the convergence speed and accuracy of the algorithm. The performance of the improved algorithm is tested on 12 benchmark functions. The experimental results show that the improved algorithm has better optimization accuracy, convergence performance and stability.

Passenger demand is the necessary foundation of urban rail transit operation management, and it’s very important to study the accuracy of short-term passenger flow prediction. However, the passenger flow is affected by many factors, so more comprehensive research is needed. This paper is based on multi-source data and the combined model. Firstly, the preliminary correlation between the passenger flow sequence and external factors is analyzed before the prediction, and the required features are extracted by using the convolutional neural network (CNN) and inputted into the prediction process. Also, the model is combined by Bi-directional Long Short-term memory network (Bi-LSTM) and attention mechanism (Attention), which can adapt to the nonlinearity and periodicity of short-term passenger flow prediction. Finally, the real example analysis shows that there is a certain correlation between subway passenger flow and multi-source data, and the CNN-BiLSTM-Attention model can improve the accuracy of passenger flow prediction and reduce errors, which is better than other traditional prediction methods.

For the geophysical logging data acquired in reality, in order to calibrate the acquired logging information, it is necessary to compare the acquired multiple logging time series data. DTW algorithm has great advantages in measuring the similarity of two time series data. However, there will be more data points "one to many" or "many to many", so the overall accuracy of curve alignment will be reduced. In order to solve this problem, this paper adopts an improved DTW algorithm. The DTW algorithm of segmented mode distance is used to correspond to the stratigraphic division. Based on DTW, the weight improvement algorithm is introduced to optimize the corresponding accuracy of the sequence and improve the matching accuracy of the corresponding points. The experimental results show that the DTW algorithm based on segmented pattern distance plays an important role in matching the changing trend of sequence features. Compared with WDTW and DTW, the accuracy of the TDTW algorithm is obviously improved, and the corresponding goal of optimizing logging data is achieved.

Clean air is a basic need for human health and well-being. Forecasting PM2.5 concentrations so that cities can make timely policy responses are of great significance of protecting citizens' respiratory health. This paper proposes a long short-term memory neural network models based on the Sparrow Search Algorithm (SSA-LSTM). The Sparrow Search Algorithm is used to optimize the number of hidden layer nodes, learning rate and number of iterations of the LSTM network to improve the forecasting accuracy of the model. In order to verify the validity of the model, the PM2.5 data onto five large cities including Beijing and Shanghai were used for time series forecasting. Compared with the existing long-short-term memory network model (LSTM), BP neural network models and bilateral long-short-term memory network model (Bilstm), the results show that the forecasting results forecast by this optimization model are the most consistent with the actual PM2.5. The forecasting error is effectively reduced, and the optimization result of the SSA algorithm has a good generalization effect.

It is difficult to track and identify shill bidding behavior brought by the development of online auctions. In order to solve this problem, this paper used the XGBoost algorithm to build a shill bidding pre-warning model and obtained the important characteristics of identifying this behavior. By comparing with other mainstream algorithms, it is found that the XGBoost algorithm has the highest accuracy in predicting the risk of shill bidding behavior, reaching 99.6%. Through the experiment and comparison of the number of features, three key indicators for identifying shill bidding behavior are found, which provides an accurate range for the attack and prevention. The research in this paper improves the identification ability of shill bidding behavior and reduces the scope of the identification characteristics of shill bidding behavior, which will effectively curb the continuous spread of shill bidding behavior.

Mining the temporal pattern and migration characteristics of population flow is of great significance for the optimal allocation of national and regional human resources, the balanced development of society and economy, and the optimization of transportation systems. This paper uses wavelet packet decomposition and reconstruction to obtain the intrinsic correlation patterns of population mobility data and combines it with the BP neural network algorithm to construct the hybrid WPD-BP neural network model. The hybrid model was utilized to predict the future migration trend, and the performance results are compared with the traditional BP neural network algorithm. The empirical results show that this paper uses WPD to provide a better data set for BP neural network, and the combination of the two shows higher accuracy and lower error rate for the same data set, which confirms the feasibility and accuracy of the method. The provided method can help the regulatory authorities to better monitor the flow of people and formulate preventive and monitoring measures reasonably according to the corresponding data changes.

Location based social network (LBSN) provides large amounts of data which record the locations visited by users and corresponding latitude and longitude of these locations. Such datasets can be used to explore visiting preferences of users and predict the locations which are likely to be visited by a particular user in the future. Thus, the problem of prediction of users’ preference locations has become a research hotspot and attracts great attention from academia and practitioners. However, it is still a challenge to precisely predict which locations will be visited by users. The main reason is that the visiting decisions made by users will be affected by not only preferences but also geographical factors. In this paper, we investigate the influence of geographical factors, and propose a gridding-based tensor decomposition algorithm for users’ preference locations prediction. We divide the entire city into grids and fill these grids with visiting records of users. A tensor is constructed according these grids, and a tensor decomposition algorithm is employed to calculate the visiting probability of each grid for each user. Then, we calculate the popularity of locations in each grid. Finally, we construct a ranking list of all locations by considering both grids' visiting probabilities and corresponding popularity scores. We have implemented our algorithm and compared with existing approaches by using two public datasets, Foursquare and Gowalla. The experimental results show that our algorithm achieves higher precision and recall.

By understanding the current situation of network security, analyzing the network security monitoring model, and analyzing the attack path and abnormal behavior of the intrusion network. The abnormal behavior of the intruder at the application and network level is summarized when the intrusion occurs, and a network security model based on multidata source information fusion is proposed according to the abnormal behavior. The model proposes five modules, which can effectively improve the network security protection ability to ensure network information security.

Satellite edge computing has attracted the attention of many scholars due to its extensive coverage and low delay. Satellite edge computing research remains focused on on-orbit task scheduling. This study explores related strategy investigation in the scenario where high-load satellites cannot participate in task scheduling based on the deficiencies of existing research work and methods. In order to solve a better scheduling scheme, a hybrid genetic particle swarm optimization (GABPSO) is proposed by introducing crossover and mutation operations of genetic algorithm (GA) into binary particle swarm optimization (BPSO). The simulation results demonstrate that the proposed method outperforms both the traditional GA and BPSO algorithms.

Most existing recommendation systems assume that items are independent and identically distributed. These methods only leverage user-item interactions as model input, while neglecting the relationship between items, leading to limited performance. To address this issue, this paper proposes the PEVGraphRec, which provides recommendations by utilizing Proximity-Effect-Value (PEV) similarity measurement to generate an item-item implicit network and employing graph neural network with attention mechanisms to intelligently aggregate the user-user connections, the user-item interactions, and the item-item relations. Specially, compared with traditional similarity measurement, the PEV method can reduce the negative effects of cold start. Experiments on three real-world datasets verify the interpretability of our model by examining the contribution of each component. The results show that the PEVGraphRec is robust and superior to other baseline methods.

In order to improve the online application ability and auxiliary decision-making ability of power grid dispatching operation rules, a method for identifying key information of dispatching operation rules based on Bidirectional Long Short-Term Memory network-Conditional Random Field (BiLSTM-CRF) is proposed. A method for feature extraction and feature labeling of dispatching operation rule text is proposed, and BiLSTM is used to learn nonlinear characteristics and implicit timing information between regular text features, and CRF is used to optimize bidirectional network encoding labels globally. The recognized rule text information is combined according to the definition level to form a knowledge graph of dispatching operation rules. Through the verification of the power grid operation rules of a control center, compared with other methods, the proposed model recognizes the average precision rate, recall rate and F1 value of 98.10%, 98.19% and 98.10%, respectively, which can more accurately support the refined retrieval and positioning of the text of the dispatching operation rules.

Dynamic graph neural networks are a research hotspot in the field of deep learning, especially in social networks, recommendation systems, traffic networks and other fields with very wide applications. At present, most of the commonly used deep learning-based network analysis methods are focused on discrete dynamic graph neural networks. However, the effectiveness of discrete graph neural network methods is based on the degree of good or bad snapshot segmentation of dynamic networks, so this paper proposes a Snapshot Clustering Segmentation Method (SCSM) for dynamic networks, and researchers can effectively segment the snapshots of dynamic networks by the SCSM method in this paper to make the downstream tasks can get good results.

Smart learning is an important means of foreign language learning in the era of education informatization 2.0. Smart Learning of Foreign Languages (SLFL) is based on students' metacognition, multiple intelligence, and meaning construction. With the support of information technology, teachers, students, learning resources and the learning environment are interrelated and interact with each other to form the framework of SLFL. Under the guidance of this framework, five paths of SLFL are constructed, namely: personalized learning, learning from crowds, learning in the real-life context, embodied learning, and blended learning, and a case study was implemented to examine the effect of SLFL.

This research aims to accomplish speech emotion recognition in Chinese by machine learning. For the training data, this paper used CASIA Chinese emotional speech database and to train the model. For the methods of classification, the researcher compares two kinds of methods including SVM (support vector machine) and CNN (convolutional neural network). In my Data preprocessing, this researcher used the MFCC (Mei frequency cepstral coefficient) From librosa to extract the feature of the audios in CASIA database. After training the feature with these two models, the researcher obtains two accuracies that are low. To improve the accuracy, the researcher tried to change the penalty coefficient and Gamma value for SVM. For CNN, this researcher tried to add drop out layer to the CNN structure and change the L2normalizer and number of epochs to increase the accuracy. In addition, due to the limitation of SVM’s classification ability, the accuracy and the performance is very difficult to improve. Therefore, this report will mainly focus on the speech emotion recognition using CNN. After the model has trained well, the researcher tested the model using the audio recorded by me and my classmate. After improving the accuracy of the model and recorded audios with more obvious identification feature, the researcher gets a very decent result.