With the rise of research on autonomous driving technology, the establishment of a scientific and complete test and evaluation system for autonomous driving algorithms has gradually attracted attention. The test and evaluation system of autonomous vehicles mainly includes testing methods, as well as the selection of evaluation plans and indicators. Different countries have also formulated standards and regulations for the development and application of autonomous vehicles, which are helpful to the establishment of the evaluation system. Therefore, this article has carried out related research work on the test and evaluation of autonomous vehicles, and introduces from four aspects:Background, testing methods, evaluation plans, different national policy and research. In terms of the background, the significance and application of the test and evaluation of autonomous vehicles are introduced. In terms of testing methods, it specifically introduces Monte Carlo simulations, game theory method, test matrix evaluation and worst-scenario evaluation. Accelerated evaluation of automated vehicles is also introduced. In the aspect of evaluation plans, this article specifically introduces the evaluation plans in Annual Research Report on Autonomous Vehicle Simulation in China, German PEGASUS project, Chinese Intelligent Vehicle Future Challenge and the European AdaptIVe project these four projects. It also introduces the evaluation plans mentioned in other academic literature. Finally, this article briefly introduces the various standards and policies formulated for autonomous driving in China, the United States, Europe and other countries and regions in recent years. Related companies and their simulation software are also introduced.

{Big data is characterized by large volume, rich types, and rapid growth, but it also has problems such as low value density and poor representativeness, which brings opportunities and challenges to sampling survey. In the context of big data, how does sampling survey adapt to new changes and what kind of development and application does it have? This paper discusses it from three perspectives. First, there are some new sampling methods with strong adaptability in the data stream environment, which can obtain representative samples efficiently and accurately, and take into account the storage space, processing time and ability. Secondly, some non-probability sampling methods without sampling frame have been developed by means of internet survey or social network data collection, which can obtain a large number of analysis samples in a short time at low cost. Third, the advantages of big data and sampling survey are integrated to integrate online and offline survey data. In the case that online sample is non-probability sample and offline sample is probability sample, this article puts forward the basic idea of data integration: On the one hand, probability samples are used to carry out the ``probability test'' for non-probability samples; on the other hand, the information of probability samples is extracted and make inferences based on model or pseudo-randomization.

This paper focuses on the preset-time bipartite consensus tracking problem of a class of heterogeneous nonlinear fractional-order multi-agent systems with signed directed graphs. By introducing a class of time-varying functions with generalized properties, a time-varying function based preset-time distributed controller is designed to achieve accurate preset-time bipartite consensus tracking in a fully distributed fashion for the heterogeneous linear and heterogeneous nonlinear fractional-order multi-agent systems, respectively. The preset time can be preset by a time-varying function, and does not depend on any initial values and parameters. Finally, one example is given to verify the effectiveness of the theoretical results.

In this paper, the technical analysis indicators are fuzz{\rm if}ied by a triangular fuzz{\rm if}ier and non-subjective transaction rules are generated using the Apriori algorithm and neural network. A fuzzy inference rule library with non-subjective categories, a product reasoning machine with the meaning of Mamdani, and a central average defuzz{\rm if}ier fuzzy Decision-making system, using recursive least squares method with forgetting factors to estimate the structural parameters of the system, and proposed two non-subjective types of trading decisions (Apriori strategy, neural network strategy). Empirical results show that after deducting transaction costs, non-subjective trading rules strategies have higher annualized returns and Sharpe ratios than passive buy-hold strategies and subjective investment strategies. Such a result can give decision makers effective investment guidance and suggestions to try to overcome their psychological and market benchmarks when fund managers or stock investors make trading decisions.

Superpixel-level hyperspectral image classification is a representative spec-tral-spatial classification method. Compared with the pixel-wise classification method, it has obvious advantages in classification accuracy and efficiency. However, the main disadvantage of superpixel-level classification algorithms is that the classification results depend heavily on the segmentation scale of superpixels. Existing literature shows that the optimal segmentation scale of superpixels is usually an experimental result, and it is difficult to be specified in advance. To weaken this dependency, a superpixel-level hyperspectral image classification algorithm based on superpixel merging is proposed in this work. Local modularity function is first used to merge the sparse weighted superpixel graph constructed. By the newly defined mapping, each superpixel is represented as a sample. Then popular KNN method is adopted to classify the merged image at the superpixel level. The superpixel merging enhances the role of spatial information in classification, effectively weakens the dependence of classification results on the segmentation scale of superpixels, and improves the classification accuracy. To evaluate the effectiveness of the method, the proposed algorithm is compared with some competitive hyperspectral image classification methods on four publicly real hyperspectral datasets. The experimental and comparative results show that the proposed method not only effectively reduces the influence of superpixel segmentation scale on the classification results, but also has obvious advantages both in classification accuracy and computational efficiency.

As the basic industry for economic development, electricity consumption is closely related to industrial structure and economic growth. Based on the panel data from 2004 to 2018, this paper empirically tests the causal relationship among industrial structure, power consumption and economic growth in Gansu Province by using panel vector autoregression (PVAR) model. The result shows that, the industrial structure and economic growth influence each other in Gansu Province. Economic growth has a positive effect on electricity consumption, and fixed asset investment throughout the country has a positive effect on economic growth and industrial structure optimization. The findings of this paper show that we should adhere to the economic development goal of steady growth and structural adjustment, and vigorously develop clean energy such as wind energy and solar energy through policy guidance. At the same time, we should also improve the efficiency of energy investment and utilization, and guide and support the development of the tertiary industry to realize the multiple goals of industrial structure optimization, energy conservation, emission reduction, and economic development in Gansu.

With the advent of the big data era, in order to improve computational efficiency, Wang, et al.(2018) proposed an optimal subsampling algorithm for logistic regression, which provides a better tradeoff between estimation efficiency and computational efficiency. To solve the problem of multicollinearity among variables, this paper proposes an optimal subsampling algorithm in the context of ridge regression, and proves the consistency and asymptotic normality of the estimator from optimal subsampling algorithm. Numerical experiments are carried out on both simulated and real data to evaluate the proposed methods. Results show that the optimal subsampling algorithm produces similar results compared with the full data analysis, while significantly reducing the computational costs.

In this paper, we extend the subsampling method under the linear model to the nonparametric regression model and propose two subsampling methods for the nonparametric local polynomial regression model. First, we derive the convergence rate of subsampling based weighted least squares parameter estimation to full sample weighted least squares parameter estimation, and the asymptotic normality of the subsample parameter estimation are derived. Then, we use the criterion of minimizing the asymptotic variance, and two subsampling methods of OPT and PL under nonparametric local polynomial regression model are proposed. Finally, numerical simulation of OPT subsampling and PL subsampling, uniform subsampling and Basic Leveraging subsampling are carried out respectively, in terms of mean square error, fitting effect and computational cost. The results show that the subsampling method based on OPT criterion and PL criterion has great advantages in improving estimation accuracy and reducing calculation burden.

With the development of information age, how to mine useful information from massive data quickly and effectively is a new challenge. As an effective tool for large scale data analysis, sub-sampling method has attracted extensive attention of scholars at home and abroad. However, the traditional sub-sampling method usually does not take into account the uncertainty of the model. When the assumed model is incorrect, the conclusions may be wrong. In order to solve this problem, a sub-sampling model averaging estimator (SSMA estimator) is constructed by the sampled data. Theoretically, we prove that the SSMA estimator is an asymptotically unbiased and consistent estimator of the model averaging estimator based on full data. In addition, we propose a weight choice criterion for the SSMA estimator, which is based on the Mallows' criterion proposed by Hansen (2007), and derive the asymptotic optimality of the weight estimator. It is worth mentioning that, in the proofs of these theoretical properties, we consider the double randomness brought by the model and sampling design. Finally, numerical analysis further shows the effectiveness of the proposed method.

This paper proposes a distributed fixed-time adaptive formation control algorithm for multiple quadrotor unmanned aerial vehicles (QUAVs) with external disturbances and parameter uncertainties via the command filtered backstepping method. By introducing the fixed-time command filter and constructing the nonsmooth error compensation mechanism, the "complexity explosion" problem is effectively avoided as well as the influence of filtered error is eliminated in a fixed time.Besides, the issue of singularity existing in the distributed controller is skillfully addressed by designing a piecewise function. Based on the fixed-time stability theory, it strictly proves that the closed-loop system is practical fixed-time stable and all the signals of the closed-loop system are bounded, meanwhile, the formation tracking errors converge to a sufficiently small neighbourhood of the origin in a fixed time. Finally, a simulation example verifies the effectiveness of the proposed distributed fixed-time control strategy.

This paper deals with the Zero-Hopf bifurcation in high dimensional polynomial differential systems. First, we reduce the problem of bifurcation analysis to an algebraic problem, and we give a method for determining the bifurcation set of the Zero-Hopf bifurcation points of differential systems by using symbolic algorithm for solving semi-algebraic systems. Then, based on the second order averaging method, the algorithmic framework of the Zero-Hopf bifurcation analysis of differential systems is derived, and the limit cycle bifurcation problem is studied through specific examples by using the methods of symbolic computation, and some new results are obtained. Finally, we propose several related research problems.

After a major disaster, victims were immediately evacuated to a safe area. The victims desperately need tents as shelters. In addition, they also need food, clothing and other materials to survive. In the absence of sufficient transportation capacity, the government have to decide which kind of materials should be sent first, and determine the deliver number of each material. In this paper we consider two critical emergency supplies, including food and tents. For one thing, to house more people, the government should allocate more transportation resources to the tents, and deliver them to the settlement as early as possible. For another, the delay in food delivery will cause suffers to the victims. To deal with this dilemma, this paper established a dynamic multi-stage emergency supplies dispatching model to maximize the number people retained in the safe area. The model restricts the victim's food deprivation level within an acceptable and reasonable threshold, therefore the transportation plan can ensure the victims' live in a certain level. Under a major earthquake setting, the paper showed a calculation example and provided a sensitivity analysis of transportation capacity and deprivation level threshold. The traditional view holds that food should be continuously transported to affected areas immediately after a disaster, however the analysis results suggest that tents should get priority in relief delivery scheduling once the victims food deprivation level could be controlled within a reasonable threshold. The sensitivity analysis results also indicate that the deprivation level threshold acts as an exogenous variable that regulates the influence of transportation capacity on the final retention people in resettlement sites. When developing relief delivery schedule, the decision makers should take both transportation capacity and deprivation level threshold into account, and adjust the victims' living standard depending on the actual situation.

In this paper, we design a kind of impulsive controller with actuator saturation to obtain chaotic synchronization of the drive-response neural network. Firstly, we use the sector nonlinear model method and the polyhedral representation method to deal with the saturation nonlinearity of the system at the impulse moment. Secondly, by selecting the appropriate quadratic Lyapunov function, combined with mathematical induction to obtain the local exponential synchronization criterion related to linear matrix inequalities (LMIs). Finally, a numerical example is used to verify the validity of the conclusion.

In order to avoid the shortcomings of the commonly used trajectory planning methods, such as the cubersome model coupling and the difficulties in the model operation, a trajectory generation and obstacle avoidance method is proposed for manipulators based on the Learning from Demonstration (LfD). This method combines Gaussian mixture model (GMM), dynamic motion primitive (DMP) and rapid extended random tree (RRT) methods after pretreating the data recorded by the robot platform. The gaussian mixture model, aiming to optimize the set of demonstration data, is employed to generate trajectories containing as many motion features as possible. DMP is used to model and generalize the movements. And then, the trajectory can be adjusted by RRT algorithm to meet the operation requirements in cases of complex environments with obstacles in different shapes. Finally, the pick-and-place experiments based on Franka manipulator validate the effectiveness of the proposed method.

In this paper, we investigate scaled group consensus problem: The agent network is divided into two sub-networks; the agent network is divided into arbitrary finite sub-networks. For these two cases, we design distributed scaled group consensus protocols, respectively, and derive the convergence states and the convergence criteria for the agents reaching the scaled group consensus by utilizing matrix theory and Routh-Hurwitz criterion. Finally, several simulations are presented to guarantee the effectiveness of the theoretical results.

The widespread spread of the epidemic has had a huge impact on economic development and daily life. Therefore, it is of great importance for formulating corresponding control strategies and economic recovery policies to collect epidemic data and analyze the spatio-temporal patterns of incidence rate or the intensity of infection. In this paper, epidemic modeling methods based on hierarchical Bayesian spatio-temporal Poisson model are discussed, including different settings of data model, process model and parameter model, discussion of parameter prior distribution, model selection and so on. Based on this idea, we can analyze the spread and development of epidemics, study the spatial differences of different regions and the influence of other covariables on epidemic trends, and study the spatio-temporal dependence of virus transmission and the heteroscedasticity structure of spatial effects.The modeling method discussed in this paper can provide theoretical reference for the study of related problems. For parameter estimation, the Gibbs sampling algorithm under the default Markov Chain Monte Carlo algorithm (MCMC) in WinBUGS and OpenBUGS can be used.

In this paper, the controllability of diffusion coupled multi-agent systems under signed networks is studied. Firstly, the upper bound of the controllable subspace of the system is given based on the generalized almost equitable partition and the restriction on the coefficient matrix of the system. Compared with the previous similar conclusions, the influence of the choice of coefficient matrix on the conclusion is discussed and a necessary condition for the controllability of the system is given: All the cells in the partition are trivial when the system is controllable. Secondly, an algorithm for computing the leader-isolated maximal generalized almost equitable partition is presented. In addition, it is proved that the controllability of the general linear diffusion coupled multi-agent system is equivalent to the controllability of the corresponding all positive network if leaders are selected from the same vertex set under the condition of structural balance, but independent of the selection of system coefficient matrix.

This paper introduces a new intelligent optimization algorithm, named Firework Algorithm (FWA), to optimize the parameter selection process in the Support Vector Regression model (SVR). And then, considering the seasonal factors in tourism economic behavior, we build a seasonally-adjusted FWA-SVR model and apply it to predict the number of overnight tourists and tourism revenue of Hainan International Tourism Island. The prediction results show that the FWA-SVR model after the seasonal adjustment has better prediction accuracy than the ARMA model without seasonal adjustment. Compared with the classic Genetic Algorithm and Particle Swarm Optimization Algorithm, the FWA-SVR model performs best in all prediction models.

In this paper, the dynamic behavior of a stochastic Ebola epidemic model with vaccination and virus transmission after death is studied. By using It\^{o} formula and constructing Lyapunov function, the existence and uniqueness of global positive solutions for stochastic systems are firstly proved; then, the asymptotic behavior of the positive solution of the stochastic system around the disease-free equilibrium and endemic equilibrium of the deterministic system is analyzed. Finally, the correctness of the theoretical results is verified by numerical simulation. The results show that environmental white noise has an effect on the stability of the equilibrium points of the epidemic model.

In order to solve the problem that the computation complexity is increased and the evaluation efficiency is reduced in functional comprehensive evaluation (FCE) caused by excessive data, this paper applies symbolic analysis technology to FCE, and proposes an interval functional clustering method based on interval functional Euclidean distance as a basic model of interval FCE. Due to the characteristics of multiple variables in comprehensive evaluation, the interval functional entropy method is proposed to construct the comprehensive index interval function. Compared with the FCE, the interval FCE adds the step of converting point value into interval data, which can better grasp the change trend of data without losing information, so as to improve the efficiency of comprehensive evaluation. Finally, the interval FCE method proposed in this paper is used to evaluate and analyze the market performance of stocks. The results show that this method has certain advantages in dealing with the comprehensive evaluation of high-frequency data, and it can be further applied in the application research.

In $x$-ray CT (computed tomography) imaging technology, there often exist the detection objects with the special sizes, shapes or materials, where the projection data can are only collected in some limited projection angles. In this case, the obtained data does not meet CT accurate reconstruction conditions. The reconstructed CT images using the conventional algorithm show some serious landslide artifacts, which are difficult to provide valuable information for many practical applications. In order to better suppress the landslide artifacts, this paper proposes the limited-angle CT image reconstruction method based on the ResNet model, in which the used residual learning method can extract the features of the input images and capture enough detail information, and then the deconvolution algorithm is used to restore the learned features. The numerical experiments prove the effectiveness of the proposed reconstruction method, which can significantly reduce the landslide artifacts and effectively retain the structural features of CT images. The reconstruction results can provide high quality CT images for clinical diagnosis.

The rapid development of blockchain token incentives provides a new perspective to solve the problem of insufficient motivation for user content creation, but it still faces many challenges in the actual implementation process. This paper takes Steemit, a knowledge community based on blockchain, as the research object.By collecting block data, this paper analyzes the situations and problems existed in the token incentive mechanism of community from two aspects, including incentive equality and knowledge contribution efficiency, thus revealing the problem of token incentive allocation monopoly. Moreover, this paper identifies the influence of token incentive allocation monopoly on user knowledge contribution. The results show that the token incentive distribution in the community is monopolized by a small number of top users, and the incentive distribution mechanism in the community cannot effectively reflect the users' knowledge contribution levels. The inequality of token incentives allocation results in the decrease of users' content production and content discovery levels.

This study examines the dynamic relationship between the US dollar exchange rate and oil prices. Based on the commodity and financial attributes of crude oil, we first theoretically analyze the pricing, settlement and portfolio effects of the US dollar exchange rate on crude oil prices. To account for the asymmetric impacts of structural break and oil financialization on dollar-oil nexus, we then embed the dollar-oil relation to a VAR framework with structural breaks and oil financialization thresholds. We further investigate the impact of investor expectations on dollar-oil nexus under the condition of oil financialization using investor self-adaptive expectation model. The results show that: First, the dollar-oil relationship has undergone significant structural change in 2001. Second, the deepening oil financialization is important reason for the structural change. The deepening oil financialization after 2001 makes the portfolio effect become the dominant effect of dollar exchange rate on oil prices. No matter in the long run or short run, the unidirectional negative causation from dollar exchange rate to oil prices is found significant. Third, under the condition of oil financialization, the impact of investor expectations on dollar-oil relation is substantial. A sharp increase in dollar exchange rate volatility largely strengthens investors' unilateral expectation to the future trend of dollar exchange rate, which will further enhance the portfolio effect, resulting in an even larger negative effect of dollar exchange rate on oil prices.

In recent years, the continuous development of social networks has improved the speed of network information transmission, so the identification of the minimum node set that can maximize the influence of public opinion information has become one of the important issues in information science. This paper integrates degree correlation and community identification, designs a method to find public opinion information source set in the DCCI social network environment, and verifies the proposed method in the network. The experimental results show that the accuracy of the proposed algorithm is slightly better than other algorithms, and the running efficiency is higher.

In this article, Fisher information in the corresponding samples about the scale parameter $\theta$ from Inverse Rayleigh distribution under simple random sampling and ranked set sampling will be respectively studied. The numerical results show ranked set sample carry more information about $\theta$ than a simple random sample of equivalent size. Then we respectively use the simple random sample and ranked set sample to construct some optimal estimators of $\theta$. The numerical results of these estimators are compared.

In this paper, problems of senor attack detection in the cyber physical system under attack signal are what we studied. Firstly, the dynamic model of the whole cyber physical system is established. Then, based on the upper and lower bound information of the whole system, two kinds of interval observers are designed under the assumption that the sensor has no attack signal. One can solve the interval observer directly under the corresponding conditions, and the other one can obtain the conditional gain matrix by linear transformation, and then convert it into Sylvester equation is used to solve the interval observer, and then the error estimation of the system output is used to determine whether there is a attack signal. Finally, a simulation example is given to verify the effectiveness of the proposed method.

In the decision-making process, how to obtain optimal group decision scheme under the premise of achieving the expected utility of individual decision-makers, is one of the main contents of multi-criteria group decision making(MCGDM). However, existing methods of MCGDM didn't consider decision-makers' expectations and risk preference for individual attributes. In this paper, we study the problem of MCGDM considering decision-makers' aspiration satisfaction. Firstly, we discuss how to apply the best-worst method to MCGDM to determine the weight of attributes. Additionally, we rank the score of alternatives by considering the aspiration and risk preference. Finally, we present a new methodology by a combination of best-worst method(BWM) and aspiration satisfaction function to solve the problem of MCGDM. We verify the feasibility of the proposed method through an example. Furthermore, we do a simulation analysis for two attributes with the largest weight.

Multidimensional systems are often described by polynomial matrices, and problems on the equivalence of multidimensional systems in system theory are often transformed into problems on the equivalence of polynomial matrices. In this paper, we mainly study the equivalence of two kinds of multivariate polynomial matrices, and obtain the discriminant conditions for the equivalence of these matrices and their Smith forms, respectively. The conditions are easily verified, and an example is also used to illustrate these in the paper.

In recent years, bilateral matching decision-making has become one of the hot issues in the field of economic management. In this paper, a new two-sided matching decision-making method for interval dual hesitation fuzzy information is proposed. Firstly, the score function of interval dual hesitation fuzzy number is defined, and the interval dual hesitation fuzzy matrix is transformed into a closeness degree matrix by using projection technique. Then, taking the maximization of the two sides undefined proximity degree and the minimization of the difference between the two sides undefined proximity degree as the goal, considering the constraints of the bilateral fair matching, a two-sided fair matching model based on the closeness degree is constructed. The ``optimal" two-sided fair matching is obtained by solving the model. Finally, taking the bilaterally fair matching problem between the production enterprise and the third party logistics company in logistics operation as an example, the effectiveness and practicability of the bilateral fair matching method based on interval dual hesitation fuzzy information bi-directional projection technology are illustrated.

As looking for high-quality online public opinions from mass reticula data is the practice of the new concept of online public opinion, both quantity and quality aspects should be considered by relevant studies. Based on the three-dimensional evaluation system of the quality of online public opinion, through the change and reconstruction of the nodes and edges of the public opinion supernetwork model, an Opinion-Noise Detection Supernetwork model was proposed, in which there are four subnetworks:Environmental subnetwork, emotional subnetwork, social subnetwork and content subnetwork. The noumenon of online public opinion “noise” refers to the public opinion data that cannot provide suggestions for the formulation of management decisions. Based on the proposed model, 18 characteristic indexes were extracted. Finally, by employing machine learning algorithms, the public opinions with high quality were successfully identified.

Complex networks have been widely used to explore the regular pattern of complex systems. This paper uses the quantile graphs method to map the daily closing price series of six stock indexes in different markets to a complex network. It analyzes the characteristics of the quantile graphs of stock indexes, and explores the changes of the network structure of stock markets in two different regions. The results show that, firstly, the network characteristics of stock index series quantile graphs in the same market are similar, but there are great differences between these two markets. Secondly, Shanghai Securities Composite Index and Shenzhen Securities Component Index have long-range correlation, and Hong Kong Hang Seng index is relatively random, but the three stock indexes in the US market, include S&P 500 Index, NASDAQ Composite Index and Dow Jones Industrial Average, are inversely long-range correlation. Finally, the quantile graphs of the two markets have different community structures. This method reveals the nature and potential dynamic behavior of stock markets in different regions from a macro perspective, and it can provide a wealth of information for stock index forecasting.

In this paper, a novel generalized functional partially varying coefficient hybrid models is proposed, which combines the generalized varying coefficient regression model with the generalized functional linear regression model. Based on the method of functional principal component base and B-splines base, the estimation of unknown functions is obtained by maximizing quasi-likelihood function, and the convergence rate and prediction accuracy of each estimator are obtained under certain regular conditions. The feasibility and superiority of the model are demonstrated by numerical simulation, and the practicability of the model is illustrated by applying the model to Tecator data.

In the recycling activities of mobile phone used electronic products, the deepening concern of users about personal privacy will affect the decision-making of closed-loop supply chain members. Based on this, by describing the users' privacy concerns and other factors, three game models based on manufacturer recycling, second-hand recycling and hybrid recycling model are constructed to study their influence on the decision-making of closed-loop supply chain members. The research shows that the increase of users' privacy concerns will promote the profits of manufacturers and retailers, but it will reduce the profits of second-hand dealers, and supply chain systems in manufacturer recovery mode are more profitable than the other two models; as user privacy concerns increase, the sales price of new products and remanufactured products (refurbishments) increase (decrease) and the sales volume increase (decrease); when users pay more attention to privacy, manufacturers can enhance their competitive advantage through privacy processing protection under certain conditions.

Locally repairable codes are a class of erasure codes which can repair multiple failed nodes. They are widely used in the distributed storage systems. A main topic in the distributed storage coding is to construct optimal locally repairable codes at present. In this paper, the following two classes of optimal ${(r,\delta)}$ locally repairable codes based on cyclic codes over $\mathbb{F}_{q}$ are constructed:1) $[3(q+1),3(q+1)-3\delta+1,\delta+2]$, where $q\equiv1(\bmod~6)$, $r+\delta-1=q+1$ and $2\leq\delta\leq q-1$ is even; 2) $[3(q-1),3(q-1)-3\delta+2,\delta+1]$, where $q\equiv7(\bmod~9)$, $r+\delta-1=q-1$, $2\leq\delta\leq\frac{2(q-1)}{3}$ is even with $\delta\not\equiv0(\bmod~6)$.

When facing high-dimensional situation, to better estimate expected return, increase the stability of portfolio strategy and obtain better out-of-sample performance, this paper uses information of double-sorted portfolio on circulating market size and book-to-market to introduce a Group-LASSO (GLASSO) regularization term into the regression-type mean-variance objective function, and constructs the GLASSO-MV portfolio strategy. Comparing to $l_1$-norm regularized LASSO-MV strategy, GLASSO-MV can effectively utilitize the pricing difference among factor portfolios, and output sparse between-group weights, attaining more effective high-dimensional weight estimation and better out-of-sample performance. To obtain suitable regularization term parameter and weight sparsity, this paper adopts 5-fold cross-validation and adjusts the sparsity based on the parameter result. In terms of empirical study, this paper uses daily data on Chinese A share market from 1995 to 2019 of 3695 stocks, and compares GLASSO-MV to multiple common portfolio strategies. The result shows that, compared to strategies including LASSO-MV, MV, GMV, TZ (Tu-Zhou), BS (Bayes-Stein), GLASSO-MV has better out-of-sample Sharpe ratio, lower standard deviation risk and turnover.

Due to the existence of estimation error, the out-of-sample performance of mean variance investment strategy is not satisfactory. At the same time, equal weight investment strategy is gradually concerned because of its lack of estimation error and good out-of-sample performance, and it plays an important role in combination strategy. Therefore, this paper introduces the equal weight strategy to adjust the original mean variance asset allocation, determines the combination coefficient of sub-strategies based on the certainty equivalence to measure the performance of different investment strategies, constructs the final combination portfolio strategy, and compares the combination portfolio strategy with other investment strategy. The results show that as far as single investment strategy is concerned, the mean variance strategy is superior in reducing standard deviation and controlling risk, while the equal weight strategy is beneficial to increase the Sharp ratio of the asset portfolio and improve return; As far as combination portfolio strategy is concerned, on the one hand, the combination portfolio strategy based on certainty equivalence can reduce the tail risk of portfolio and help investors avoid extreme losses, on the other hand, it can significantly improve Sharpe ratio and Sortino ratio of portfolio and obtain higher risk-adjusted returns.

A model selection method for the higher-order spatial autoregressive model is studied in this paper, which can select the spatial weight matrix and covariate simultaneously. The asymptotic efficiency of model selection is proved in the sense of Kullback-Leibler loss. Further, we propose a model averaging method for the higher-order spatial autoregressive model, and its asymptotic optimality is established. The numerical simulation results demonstrate the merits of the model selection method and also show that the model averaging method can further improve the performance of model selection.

To handle the tracking control problem of unknown strict feedback nonlinear control systems with dead zone input, a projection-based adaptive command filtered and finite-time control strategy is proposed based on immune function. The immune function and the extended state observer are utilized to estimate the unknown information of control systems with dead-zone input. The command filtering is used to cope with the derivative explosion problem occurred in the backstepping method and the error compensation mechanism is established to diminish the impact of the filtering error on tracking accuracy. The projection operator ensures the boundedness of adaptive parameters. Compared with the literature on adaptive backstepping constrained control based on a barrier Lyapunov function, this paper simultaneously guarantees system states, the compensated tracking errors and adaptive parameters within pre-assigned certain bounds and ensures the boundedness of all closed-loop system signals. It enhances the controller’s convergence rate by using the finite-time control. Finally, a simulation example shows the effectiveness of the proposed control strategy.

To accurately identify the head defects of the GH159 bolt after hot upsetting, this paper proposes a defect recognition method based on transfer learning, where datasets under scenes with different brightness are set as the source domain and target domain in transfer learning, respectively. First, considering the multi clusters of the conditional distribution in the domain, this paper adopts the K-means algorithm to cluster samples with the same defect and determine the cluster centers in this defect, then a novel measurement of the distribution discrepancy can be constructed on the cluster centers. Second, based on the distances between cluster centers and the distances between each cluster center and the samples belonging to the cluster, a new intra-class discrepancy can be established for improving the computational efficiency of transfer learning. Finally, the optimization objective of the proposed method is built on minimizing the weighted sum of the constructed distribution discrepancy and intra-class discrepancy to effectively identify defects under scenes with different brightness. According to the requirement on partial parameters setting of the proposed method, the pseudo-accuracy is designed using the reverse verification strategy, then the parameters are set as the parameters’ combination with the highest pseudo-accuracy. Using the collected dataset on head defects of the GH159 bolt after hot upsetting, the analysis and application of the defect recognition are carried out to verify the effectiveness of the proposed method.

Liver cancer has the second highest fatality rate among all cancers. Machine learning methods can improve the accuracy of disease prediction. Therefore, in this paper we mainly apply machine learning methods to study the pre-diagnosis problem for liver cancer, and improve the prediction accuracy to liver cancer. Firstly, 10 indicators affecting liver cancer are selected as predictors, and 579 liver cancer patients are divided into two groups:A training sample composed of 492 patients are randomly selected, and a testing sample composed of the remaining 87 patients. Then, we take advantage of the training samples to establish six classifiers:Logistic regression, $L_{2}$ penalized logistic regression, Support Vector Machine (SVM), Gradient Boosting Decision Tree (GBDT), Artificial Neural Network (ANN) and eXtreme Gradient Boosting (XGBoost), where logistic regression and $L_{2}$ penalized logistic regression adopt Newton-Raphson algorithm to obtain the iterative weighted least squares estimators for model parameters, calculate the probability estimate of malignant and benign tumor cells in patients, and determine the optimal threshold to predict tumor traits. Finally, the confusion matrix, sensitivity and specificity are calculated by the testing samples, and the ROC curve is drawn to evaluate the prediction accuracy. The results show that in terms of prediction accuracy, $L_{2}$ penalized logistic regression ranks the first, SVM prediction accuracy ranks second, XGBoost prediction accuracy ranks third, logistic regression prediction accuracy ranks fourth, GBDT prediction accuracy ranks fifth, and the prediction accuracies for ANN and random forest are the worst.