This paper establishes a multi-objective optimization model for the process of yoga course ideology and politics in colleges and universities through the group paper technology, and adopts an improved genetic algorithm to solve the model. Integrating the online and offline hybrid teaching mode, the paper’s group paper technology is placed in the online assessment to realize the all-round reform of yoga course Civics teaching. The research results show that the improved genetic algorithm (IGA) in this paper has a higher grouping efficiency and quality compared with the traditional genetic algorithm (GA), and saves about 52.7% in the grouping time. At the same time, using the social network analysis method to analyze the online-offline hybrid teaching mode of this paper, we are able to derive the student objects that need to be focused on based on the results of the analysis of the centrality of the point degree and the centrality of the middle of the point degree. The experimental class adopting the teaching mode of this paper has an excellent class performance rate of 100% under the test of group paper technology, which fully demonstrates that the teaching reform method proposed in this paper for the ideology of yoga course in colleges and universities has significant practical application effects.

This paper constructs an overall framework for vulnerability mining, covering the whole process from code collection to vulnerability remediation. The word vector technique is used to transform code fragments into vector form, thus preserving the semantic information of the code. A vulnerability mining system based on semantic graph of source code is further designed, which generates a semantic graph of code by constructing an abstract syntax tree (SAT), and analyzes the semantic graph by using graph neural network to accurately locate potential vulnerabilities. At the same time, a vulnerability repair method based on thought chain is proposed. The results show that the model in this paper can accurately mine the vulnerabilities of web service software, and it consumes short latency and has strong stability. The results of web service software vulnerability detection show that the accuracy rate of the model always stays above 85% under different network structures. In addition, this paper obtains that the integration degree centrality measure and 60 iteration rounds have the best effect on the detection of vulnerabilities of the model. Finally, the vulnerability repair experiments show that at Beams=15, the model in this paper repairs each vulnerability function with a PPP metric of 61.52% and an average time of 3.168 seconds, which is the best for vulnerability repair.

Because of their inherent vulnerability, small States are in a relatively weak position in global economic relations. In order to survive and develop, regional cooperation is an important way for small countries to safeguard and promote their national interests. Taking Malta as an example, this paper utilizes the international trade data of some major countries along the “Belt and Road” from 2004 to 2018, and adopts the social network analysis method to construct the trade network of 37 countries in each year, and explores the formation of the trade network of the “Belt and Road” through the gravity model. The gravity model explores the major trade subgroups formed in the “Belt and Road” trade network and reveals the cooperation and interaction paths of small countries in the global economic system. Among them, Malta’s position and role in the countries along the “Belt and Road”, Malta still participated in global economic cooperation through Russia in 2004, and in 2018, in the more closely linked global economic network system, Malta strengthened its economic ties with China, which has a core degree of 0.897, further enhancing its own position in the global economic network. The economic scale and institutional differences between countries have different impacts on the trade of different products, and small countries have limited resources, so it is more important to formulate different economic development strategies according to different influencing factors and product types, and strengthen trade cooperation with neighboring countries.

The professional development of physical education teachers is the direction and basic requirement of modern education development, which is better promoted by strengthening the tracking and prediction of the trajectory of the professional development of physical education teachers. In this paper, a combined ARIMA-LSTM model is established to visualize the PE teachers’ professional development trajectory by predicting their professional development scores, using the advantages of ARIMA model in handling linear time series data, while combining the powerful ability of LSTM network in capturing the long-term dependency of data. Three physical education teachers were randomly selected as research subjects to predict their PE teacher professional development trajectories. The root mean square error (RMSE) and mean absolute percentage error (MAPE) values were used as the assessment indexes of the model, and the MAPE and RMSE of the ARIMA-LSTM model were less than those of the ARIMA model and the LSTM model for the physical education teachers in No. 1 and No. 3. And on the prediction of physical education teacher No. 2, the MAPE comparison of ARIMA-LSTM model and LSTM model increased by 1.12%, but the RMSE decreased by 73.4563, and the prediction curve of the professional development score was close to the original sequence, and the ARIMA-LSTM model still showed better prediction effect.

In order to improve the accuracy of enterprise financial risk early warning and realize the risk control of enterprise capital chain under digital transformation, this paper adopts the Long Short-Term Memory (LSTM) neural network algorithm to establish the enterprise financial risk early warning model. First analyze the enterprise financial risk early warning indicators, use factor analysis for indicator screening, determine the indicator weights through the Delphi method and the improved hierarchical analysis method, and select the indicators with high importance to construct the enterprise financial risk early warning feature samples. Then after LSTM neural network training, the enterprise financial risk early warning model is obtained, and the model is evaluated for performance and practical use. The experiment proves that the accuracy of the LSTM neural network model on the training set and test set is 91.48% and 88.62% respectively, which indicates that the model can effectively predict the enterprise financial risk. By comparing with the commonly used enterprise financial risk warning algorithms, the algorithm in this paper has the highest warning accuracy, shorter prediction time, and better warning performance in dealing with large-scale enterprise samples. This study provides an effective financial risk early warning method for enterprises, which can help them better carry out capital chain wind control in the process of digital transformation.

Vocal music is an art about the perception and expression of sound. Successful vocal music teaching is to cultivate students’ unique singing personality, so this paper constructs a personalized vocal music teaching mode with the help of AI algorithm. Subsequently, it describes the problem of service object learners under personalized learning path recommendation, proposes a personalized learning path recommendation strategy based on ant colony optimization algorithm, and verifies the recommendation effect of personalized path through simulation experiments. Then the cognitive diagnosis model based on KM-VDINA is proposed to diagnose students’ vocal music knowledge under personalized learning path. The article concludes through experiments that the personalized vocal music teaching model based on AI algorithms requires the integration of online and offline teaching, while focusing on the integration of teaching inside and outside the classroom. The vocal music learning path of most students can be expressed as (000000)→(100001)→(101001)→(101100)→(111100)→(101110)→(111111). Students have multiple trajectories to master the attributes of vocal music knowledge, so teachers can explain the attributes of knowledge that are easier to master according to the actual situation, and then explain the attributes of knowledge that are difficult for students to master.

The purpose of this paper is to study the multidimensional computational optimization problem of enterprise organization structure. Based on the value dimension, this paper designs a kind of enterprise organizational structure which contains six dimensions such as demand deϐinition, sales and so on. And the optimization model of enterprise organizational structure is studied by taking a demand-pull enterprise as an example. In the iterative process of the algorithm, the particle swarm optimization (PSO) algorithm and ant colony optimization (ACO) algorithm are run alternately to form the PSACO algorithm to solve the optimization model of enterprise organizational structure constructed in this paper. The experimental results show that the PSACO algorithm used in this paper can maintain stable and good convergence performance under different demand sizes. The enterprise organizational structure optimization model and solving algorithm in this paper can achieve ideal resource allocation scheme and effect, and can still achieve good solving effect in large-scale enterprise organizational structure optimization adjustment. It provides good decision support for the optimization and improvement of enterprise organizational structure in the era of digital economy, and has important application value.

In this paper, we first constructed a mathematics network course goal achievement index system with 5 primary indicators, 16 secondary indicators and 70 evaluation points to provide a scientific basis for course design. After that, based on the radial basis function (RBF) neural network structure, the fruit fly optimization algorithm (FOA) is introduced to dynamically optimize the parameters of the RBF model, and the dynamic FOA-optimized RBF neural network model is put forward to predict the degree of achievement of the course objectives. The results show that the model in this paper has good convergence and prediction accuracy, and its error on the four course math network goal attainment is only about 0.4%, with excellent model accuracy and simulation effect. Combined with the prediction results, considering the shortcomings of the current teaching, a blended teaching model based on mathematics majors is constructed, and the advantages of the teaching scheme in this paper are elaborated, which provides support for the teaching reform of mathematics courses.

With the development of transportation systems, there is an increasing demand for real-time understanding of traffic scenes using image segmentation algorithms. Therefore, this paper carries out an in-depth study on how image segmentation algorithms for complex traffic scenes can meet the detection requirements of real-time while maintaining accuracy. The article first proposes a lightweight semantic segmentation method based on IDEL_DeepLabV3+, which lightens the IDE_DeepLabV3+ network and optimizes the loss function to improve the positive and negative sample imbalance problem. Then an improved image multi-texture detection method based on Faster RCNN is proposed to improve the detection performance of complex traffic scenes. Finally, the performance of the algorithm designed in this paper is tested through experiments. The performance of the deformable convolution, attention mechanism and feature pyramid improved model is tested and verified, the AP value of the deformable convolution is increased from 41.36 to 47.26, the mAP value of the overall model of the scSE attention mechanism is increased by 0.84%, and the final AP value of the weighted bi-directional feature pyramid network reaches 45.4. The improved DeepLabv3+ network achieves a high AP value of 75.03% in terms of the evaluation index mIOU by 75.03% is better than the original network’s 72.26%, so it can be said that we experimentally verified that our improved method enhances the segmentation accuracy of DeepLabv3+ network. The experimental results show that the proposed method in this paper improves the image segmentation accuracy while guaranteeing the segmentation speed, which effectively improves the segmentation effect.

Traditional non-heritage music art is gradually declining, driven by cultural heritage education, this paper studies the future inheritance trend of non-heritage music art. In this paper, the infectious disease dynamics model and complex network propagation theory are used to construct a mathematical model of the inheritance of non-heritage music art, an improved non-dominated sorting genetic algorithm is proposed, and the trend characteristics of the inheritance of non-heritage music art are simulated by solving the parameters of the model through the improved NSGA-Ⅱ algorithm which introduces the congestion calculation method and the crossover strategy. The improved NSGA- Ⅱ algorithm shows better convergence speed of optimization search and uniformity of solution distribution on single peak function and three ZDT functions. Its SP and IGD indexes are much better than the comparison algorithm, with values less than one-half of the comparison algorithm. Taking the heat of non-heritage music and art inheritance in Baidu index for several days as the simulation object, it is found that the simulation of the mathematical model of non-heritage music and art inheritance in this paper has a maximum heat value of 115,000 and the real maximum heat value of 117,241 are not much different from each other, which confirms that this paper’s non-heritage music and art inheritance mathematical model has a better fitting effect and reasonableness. The work of this paper has injected new vitality into the innovation of non-heritage music art inheritance.