Least Squares Support Vector Regression (LSSVR) machine has the advantages of small sample, nonlinearity and high dimensionality, which can solve the problem of predicting the compressive strength of green concrete with oil palm shell aggregate. In this paper, the error sum of squares instead of the error sum is used as the objective function, IFFA is used to find the optimization of the kernel function parameters and penalty factors of LSSVR, and the PWLCM-based chaotic search is used to initialize the population, and ultimately the improved auricular fox algorithm is realized for the optimization of the least squares support vector regression algorithm, which makes it have strong fitting and generalization abilities, and significantly reduces the burden of computation, thus improving the Computational efficiency. Application of the designed combined algorithm for compressive strength prediction of concrete reveals that the R², MAPE and RMSE values obtained by this paper’s model on the training dataset are 98.71%, 5.92% and 1.0823 MPa, respectively. The correlation coefficients predicted by the model are much closer to 1 as compared to that of the baseline model, which suggests that this paper’s model possesses a superior generalization capability, making it more effective in dealing with complex and invisible data. The adopted method is practical and innovative, and is of guiding significance for practical engineering.

Starting from the essence of dynamic programming algorithms, the terminology in dynamic programming algorithms, the applicability conditions of the algorithms, and common sub-problem models are summarized. The Belman optimal algorithm is used to split the multilevel problems in dynamic planning into simple single-level problems and solve them one by one, combined with the function approximation structure to approximate the performance index function, to construct the adaptive dynamic planning algorithm, and to apply it in the resource balancing optimization of integrated teaching. The results show that the adaptive dynamic programming algorithm has better resource balancing effect than other algorithms, and the number of convergence and running time are reduced by 6-53 times and 48.92-90.34 seconds respectively. The introduction of the adaptive dynamic programming algorithm improved the resource balancing accuracy of university teaching and learning management by 4.0%-17.4% in each subject group. As the number of resources increased, the time consumption required when balancing resources decreased by 50%-83.33% for test groups 3, 4 and 5, and the efficiency of the test improved by 75%-100%. This shows that the algorithm proposed in this paper is effective when dealing with balancing online and offline teaching resources in higher education.

Grasping economic behavior is a foothold of market optimization, this paper combines game theory ideas with deep learning technology to explore the laws of economic behavior. Firstly, different payment strategies are considered and payment matrices are constructed, and replicated dynamic equations are used to describe the dynamic adjustment process of the game and simulate the game process of market economic transactions. The MS-RCNN model that can be used to predict economic behavior is constructed by extracting data features using CNN and processing the feature matrix using GRU. The results of the game simulation simulation show that when the government regulation of the market is in place, it is easier for the trading parties to reach a deal. In addition the MS-RCNN model can more accurately reflect the fluctuation of the market when making long-term and short-term predictions, and the predicted price is closer to the real market price. Therefore a better understanding of economic behavior through game theory and its prediction through deep learning helps to achieve the optimization of market strategies.

The change of landscape pattern is closely related to the quality of ecological environment, and the study of urban and rural landscape pattern, especially three-dimensional landscape pattern, is of great significance for urban-rural integration spatial planning. Based on the theory of landscape pattern, this study constructs a numerical simulation method for the characteristics of urban and rural threedimensional landscape pattern, and explores the formation of optimization strategies for the threedimensional development of urban and rural areas. Taking Chengdu City as an example, firstly, based on multi-source remote sensing data, the landscape pattern index method and gradient analysis method are utilized to explore the spatial and temporal coupling characteristics of urban and rural three-dimensional landscape patterns. Then the CA-Markov coupling model is used to predict the landscape pattern of future land use, so as to provide a reference for decision-making. The results of the study show that the landscape type changes in Chengdu City from 2005 to 2020 are dominated by the transformation between cultivated land, forest land and construction land, and the reasons for the changes are closely related to the urban development plan. In addition, the accuracy indices of the CAMarkov model all reached more than 80%, and the simulation results were reliable. The model prediction results show that construction land and cropland are the largest transformed landscape types, with a large-scale increase in the landscape area of construction land and a large-scale decrease in the landscape area of cropland. Spatially, the degree of fragmentation of the landscape pattern in Chengdu City gradually decreases, the landscape patches are more regularized, and the overall pattern shows a highly aggregated trend. The research results of this paper can be used as a reference for the optimization policy of three-dimensional landscape pattern in urban and rural areas, and provide data support and innovative ideas for the innovative development of urban and rural three-dimensional landscapes

As a core course of Business English majors, Business English translation plays a crucial role in the cultivation of Business English talents, and how to realize the assessment of translation efficiency in teaching has become a hot topic nowadays. This paper builds up a translation efficiency assessment index system in the teaching of business English translation around five aspects: vocabulary, syntax, context, society, and translator’s factors. Random forest and Lasso regression methods were used to select 15 feature variables including sentence order and collocation between words. The multiple regression linear model was chosen to construct a model for assessing translation efficiency in business English translation teaching, and the model was estimated and tested. The least squares method was used for estimation, and all the parameters were significant (Sig<0.05) except for the variables compound sentences, sentence structure and situational intermingling. The distribution of the residuals of the model approximates to the normal distribution, which satisfies the assumption of normality and the assumption of independence, and possesses a good fit and some explanatory power.

This paper optimizes the K-means clustering algorithm based on the RFM model improved by the entropy weight method and then using the distribution between the samples, and adopts the combination of both density and distance to accurately classify the cross-border e-commerce customers. Finally, the capsule network recommendation model is used as the benchmark model, and the CCN4SR model is designed to accurately recommend goods to customers. The results show that cross-border e-commerce customers are categorized into five-star to one-star customer groups, which focus on “return on investment, pursuit of social value, the pursuit of cost-effective, the pursuit of low prices, while having their own different consumer preferences”. The capsule network outperforms CNN on both training and test sets, and its precision, recall and F1 value are above 92% on the test set, which shows that the capsule network is well adapted in the ϐield of implicit feedback recommendation.

There are many mature traditional navigation algorithms, but most of them are insufficient in the function of environment perception and understanding, and reinforcement learning can give robots the ability to learn and make decisions. This paper proposes a robot reinforcement learning navigation algorithm and optimal control strategy based on deep reinforcement learning. Firstly, Markov decision modeling for local planning of the robot navigation system is implemented, and then a POMDP belief space dimensionality reduction algorithm based on the NMF update rule is proposed to address the situation of excessive dimensionality and combined with PRM to achieve global reinforcement learning planning. Finally, considering the external information interference problem, a power controller based on the TD3 algorithm is designed to ensure that the robot navigation system can accurately track the signals even under the external interference environment.The position error of the robot under the TD3 controller tends to be close to 0, which is much lower than that of the robot under the PD controller. The experimental results of this paper show that the designed TD3 controller can effectively improve the trajectory tracking accuracy of the robot navigation system and better realize the optimization of the robot tracking control function.

Cataract, as an extremely common visual impairment disease, seriously affects the normal work and life of patients, and the optimization of cataract IOL model is of extraordinary significance to the diagnosis and treatment effect. The article collects ocular biological measurements of cataract surgery patients as experimental data, explores the radial basis function (RBF) neural network belonging to the field of artificial intelligence in the process of IOL calculation, and then introduces genetic algorithms to optimize the RBF neural network, and constructs the cataract IOL calculation model based on GA-RBF. The experimental results show that after combining the improved cataract IOL calculation model for telemedicine, the patient’s hospitalization days were shortened by 3.06 d, and the hospitalization cost decreased by $1,383.7, meanwhile, the patient’s satisfaction increased by 4.69%.

This paper investigates and analyzes the optimal allocation of educational resources and the expansion and innovation of the content system of e-commerce English courses in vocational education institutions in Fujian and Taiwan, and proposes methods and strategies for the optimal allocation of educational resources and the innovation of the course system. The evaluation index system of educational resources allocation was established, the factor analysis method was used to establish the educational resources allocation measurement model of vocational colleges and universities, and the K-Means clustering algorithm introducing profile coefficients was applied to cluster vocational colleges and universities on the level of educational resources allocation. The study classified 42 vocational colleges in Fujian and Taiwan into four categories, and based on the results of cluster analysis and factor ranking, the four categories of vocational colleges put forward suggestions for optimizing the allocation of their own educational resources allocation level. The results of the curriculum system innovation practice show that after the teaching design of the e-commerce English curriculum system innovation, the performance of the experimental class is significantly higher than that of the control class, increasing from 22.84 to 26.81 points. It shows that the teaching design of ecommerce English course system innovation is suitable for the needs of English teaching and can provide important guidance for teachers of e-commerce English in vocational colleges and universities when they are teaching.

This paper combines the multifactorial influence of the actual situation, adds the objectives of user interest preference and traditional music overseas communication budget into the influence maximization model, and constructs the Multi-Objective Influence Maximization Model (MOIM) of Chinese traditional music overseas communication to deal with the problem of objective inconsistency in the process of music communication. After that, the seed node selection algorithm of MOEA/D based on decomposition strategy is proposed to improve the search optimization strategy of seeds in the MOIM model. The cross-variance operator designed in the algorithm optimizes the set of solutions generated by the chromosome in the iterative process and finally obtains the Pareto non-dominated solution. The results show that the distribution of Pareto optimal solutions for each graph in the three datasets of TFM, TCC and TCO is very uniform when T=300, and the distribution of Pareto optimal solutions is more uniform with the increase of the number of iterations. The more influential nodes in the multi-objective optimization model of this paper, the higher the cost. The influence and cost of the seed set need to be considered in the overseas dissemination of music, and the seed set should be selected to maximize the influence within the budget. When the network structure and user behavior conform to different characteristics, the MOEA/D model can also get the corresponding undominated solution.The MOEA/D model integrally optimizes the influence index and cost index, so it provides a more flexible set of decision-making solutions for the overseas dissemination of Chinese traditional music.