[Retracted] Evaluation of the Modern Value of Imperial Examination Culture in the Context of Cultural Confidence Based on Deep Learning Models

Zhao, Xiu

doi:https://doi.org/10.1155/2022/9748146

Mobile Information Systems

On this page

Abstract Introduction Related Work Analysis Conclusion Data Availability Conflicts of Interest Acknowledgments References Copyright Related Articles

Research Article Retraction

!

This article has been Retracted. To view the article details, please click the ‘Retraction’ tab above.

Special Issue

Federated Learning, Internet of Things, and Edge Computing for Smart Services

View this Special Issue

Research Article | Open Access

Volume 2022 | Article ID 9748146 | https://doi.org/10.1155/2022/9748146

[Retracted] Evaluation of the Modern Value of Imperial Examination Culture in the Context of Cultural Confidence Based on Deep Learning Models

Xiu Zhao¹

Academic Editor: Muhammad Zakarya

Received03 Jul 2022

Revised13 Aug 2022

Accepted17 Aug 2022

Published12 Sept 2022

Abstract

National prosperity, national spiritual independence, and cultural security greatly impact the cultural self-confidence. It is also a new era issue that must be answered in building a socialist cultural power. In this paper, we scientifically define the core concepts of the imperial examination system, imperial examination culture, and imperial examination cultural heritage. We systematically discuss the connotation and extension of the imperial examination cultural heritage, analyze the positive energy and negative effects of the imperial examination culture, propose the construction of a new type of examination culture system, and innovate the content and form of the examination. We conduct an in-depth study on the modern value of the imperial examination culture in the context of cultural confidence and then construct appropriate evaluation indicators to use the deep learning model to evaluate the value. The technical principle and performance optimization method of DBN neural network are introduced and then the modern value evaluation system of imperial examination culture is constructed. The parameters of the DBN model are optimally selected through experiments. After the optimal model is trained, the sample data are input to obtain the results. The results are compared with the evaluation results of human experts to judge the performance of the model. In many cases, the results are very close to, or the same as the expert results, in some of the experiments, the accuracy of our model varied from the expert results by a small margin of up to 0.4%.

1. Introduction

The rise and fall of national prosperity, cultural security, and national spiritual independence all have a significant impact on cultural self-confidence. It is also a new era issue that must be answered in building a socialist cultural power [1]. The imperial examination system was first developed in the Sui and Tang Dynasties, perfected in the Song and Yuan Dynasties, and flourished in the Ming and Qing Dynasties. It lasted for 1,300 years in Chinese history. It had a major impact on the politics, economy, education, psychology, behavior, and culture of ancient China and the world, and it was a major institutional creation that shaped the cultural self-confidence of China, the Chinese people, and the Chinese nation. It is known as one of the most distinctive features of Chinese civilization, and it constituted a strong institution. The system was originally developed to recruit officials for the imperial government and consisted of two systems, the imperial civil service examination system, and the imperial military examination system. The imperial examination system is not only a political system but also an educational examination system, which is closely related to education [2, 3]. Cultural self-confidence is a more fundamental, larger, and longer lasting kind of self-assurance as well as a core soft power. The examination system represented by the imperial examination system is a grand event for the country to select talents. It is the traditional Chinese examination culture that has been nurtured in the process of civilization development for more than 5,000 years. The innermost spiritual aspirations of the Chinese people have been gathered by the cultural genes of the examination, which include self-advancement via investment, selection of academics from the brilliant texts, and seeking talents for the nation. It represents the most unique examination concept logo in China, reflects the most basic psychological and behavioral characteristics of the Chinese people, and is an important part of strengthening the cultural self-confidence of the Chinese people [4–7]. The imperial examination culture and modern examination culture have played a huge role in promoting economic development and social progress. Only by deeply understanding this logical starting point and its modern value can we better realize the creative inheritance of imperial examination culture and the innovative transformation of examination culture and continuously contribute Chinese examination wisdom and Chinese examination strength to the progress of human examination civilization. The new examination culture system is the critical inheritance, creative transformation, and innovative development of the excellent part of the imperial examination culture [8]. The imperial examination system is the cultural soft power of ancient China, and the college entrance examination is the cultural soft power of modern China. The socialist examination culture system with Chinese characteristics built on it is formed and developed under the leadership of the party. In the unremitting struggle of socialist selection of builders and successors for the comprehensive development of morality, intelligence, physique, beauty, and labor, they have taken root in the vast land of China, absorbed the examination and cultural nutrients accumulated by the Chinese nation's long struggle, and learned from human civilization about talent selection. The outstanding achievements of the college, especially the college entrance examination system and culture, are in line with China's national conditions and have a profound practical foundation and broad development prospects [9]. The imperial examination system is a major event in Chinese history, and the imperial examination culture has become one of the basic components and core elements of Chinese traditional culture, so it is worthy of continuous exploration from a single-disciplinary, multidisciplinary, or interdisciplinary perspective [10]. Imperial examination culture is the source of the discipline culture of imperial examinations, which is a specialized research field or specialty that studies the imperial examination system and its operation history that existed in the history of China and other East Asian countries. It is also an applied interdisciplinary subject. The research method is to study the emerging borderline interdisciplinary subjects based on the characters, events, and the operation law of the elements of the imperial examinations. The formation of the disciplinary culture of imperial examinations is closely related to the construction of imperial examinations. The imperial examination culture is the theoretical source of the cultural construction of the examination discipline, and it is also the most basic part of the thinking, knowledge, framework, and methods of the examination discipline [11]. Examination subject culture belongs to subculture in the examination culture system, and its formation is a sign of the maturity of imperial examinations subjects. According to the principle of academic general agreement, subject culture can promote the high-quality development of the dominant subject areas of examinations and also influence the development trend of examination subjects and the evolution of the frontiers of examination subjects [12]. Imperial examination culture is the internal motivation for its research field to develop into subdisciplines, related disciplines, and mutual integration to form interdisciplinary and borderline disciplines. It is an important aspect that should be strengthened in the construction of its discipline culture and ultimately promote the scientific, standardized, and institutionalized disciplines of imperial examinations and examinations through the power of discipline culture [13, 14]. As a great institutional invention of our nation, the imperial examination system is highly praised by foreigners and contains extremely rich institutional cultural resources and treasures of wisdom. For example, the spirit of openness, fairness, and justice manifested by the perfect examination management system and strict punishment legislation. This undoubtedly has universal enlightenment and reference significance for fair and reasonable system construction in various fields today. In addition, the imperial examination system, as an important examination system for selecting officials in ancient times, also has a mirror value for today's examination reform. Modern examinations, especially large-scale educational examinations, also undertake complex and heavy social functions, and their waste has the characteristics of affecting the whole body [15].

We should earnestly learn from experience and lessons from the abolition of the imperial examination system, and we should consider the examination reform carefully and proceed with caution. This paper conducts in-depth research on the modern value of imperial examination culture in the context of cultural confidence and then constructs appropriate evaluation indicators to use deep learning models for value evaluation. The main contributions of the paper are the following.

We define the core concepts of the imperial examination system, imperial examination culture, and imperial examination cultural heritage and systematically discuss various aspects of imperial examination culture that include connotation and extension of the imperial examination cultural heritage, the positive energy and negative effects of the imperial examination culture, and the modern value of the imperial examination culture. We introduce the research progress of imperial examination culture and then propose the construction of a new type of examination culture system, consolidate the cultural foundation of the examination subject, and innovate the content and form of the examination. The study introduces the technical principle and performance optimization method of DBN neural network and then constructs the modern value evaluation system of imperial examination culture. The parameters of the DBN model are optimally selected through experiments. After the optimal model is trained, the sample data are input to obtain the results. A comparison of the achieved results with the expert’s evaluation results shows that the error rate of our model is very low and the results are close, or the same as those of the expert’s evaluation.

Since the 1980s, there has been an upsurge of cultural discussions in the ideological and cultural fields of our country for more than ten years. The research on the imperial examination system has ushered in a spring of blooming flowers and entered a period of continuous upsurge. Not only are monographs published every year, but the number of published papers has doubled, and graduate students have chosen imperial examinations as their thesis topics [16]. Due to the emancipation of the mind and the relaxation of the academic environment, the research on the imperial examination system was unprecedentedly active. Reference [17] pointed out that the imperial examination system was a consistent system in 1300. As an examination system, it is unique in the world because of its early implementation, long duration, and great influence. Due to the factors of the imperial examination system itself, he had tenacious vitality in the feudal society. The historical role of the imperial examination system in its emergence, completion, and prosperity should be affirmed. Reference [18] argues that the historical role and status of the imperial examination system should be reunderstood, and the functions and specific contents of the imperial examination system should not be confused. As far as the system itself is concerned, the imperial examination system is worthy of being a masterpiece of Chinese traditional culture and has played an extremely important role in Chinese history. Like other essences of traditional culture, the positive parts of the imperial examination system are worthy of our inheritance and development. Reference [19] argues that the imperial examination integrates various functions such as cultural inheritance, educational supervision, value orientation, resource allocation, and social regulation, but the most far-reaching influence is the educational value orientation. The supervising function of the national selection examination in the realization of its training objectives in education cannot be eradicated as a malady. After the 1990s, while continuing to strengthen the examination and interpretation of the system, many scholars began to pay attention to the theoretical study of the imperial examination system. Reference [20] made an in-depth and original discussion on the Qing Dynasty champion from a new perspective. The author uses the method of Madhyamika historiography to study a period of the imperial examination system and a group of relevant figures and draws many valuable and convincing conclusions.

In the 21st century, the research on the imperial examination system has continued to heat up, and the research on the imperial examination system has entered the disciplinary stage of “Imperial Examination Science” [21]. Taking the imperial examination system and its operation history existing in China and other East Asian countries as the research object, and around this research object, research has been carried out at multiple levels in the research field, forming a relatively stable research scope and content. There are those who conduct overall research on the imperial examination system, and some who conduct research on the imperial examination system in different generations [22]. More importantly, the research on the imperial examination system has shown a multidisciplinary research trend, and there has been a multidisciplinary research on the imperial examination system in political science, sociology, cultural studies, education, and literature [23]. In recent years, the diversification and integration of research methods of the imperial examination system have been outstanding. In addition to comprehensively using the methods of comparison, connection, analysis, induction, and verification, many studies also generally use statistical and quantitative analysis methods to study the geographical distribution, geographical movement, social mobility, and the scale of examinations at all levels and their acceptance rate and other issues [24]. Reference [25] published a large number of important papers on the imperial examination system and successively undertaken several national-level projects related to the imperial examination system, which greatly enhanced the academic awareness of the imperial examination system research and the academic value of scientific research results. There is an inseparable relationship between the imperial examination system and literature, and its influence on literature is also multifaceted. From the perspective of research content, scholars' research mainly focuses on examination-style research, literature research, and Baguwen research [26]. With the deepening of the research on the imperial examination system, the cultural research of the imperial examination system has gradually attracted the attention of the world. Reference [27, 28] expounds the origin of the imperial examination system, its relationship with Chinese culture in its historical evolution, and its influence on the scholar community. On the basis of fully absorbing the research results of predecessors, it digs deep into historical materials and uses a combination of vertical and horizontal methods. The research method longitudinally examines the course of the reform of the imperial examination system and horizontally explores the cultural factors of the decline and fall of the imperial examination system. Reference [29] is a comprehensive perspective of the imperial examination system from the perspective of “big culture,” which opens up new horizons. People tried to integrate the “Imperial Examination System” and “Culture” together to form a new research field of “Imperial Examination Culture.” Reference [30] pointed out that the imperial examination culture as a whole culture, in a broad sense, refers to the concepts, systems, and material forms of culture related to the selection of talents by subject examinations. It is guided by the political concept of “uniformity,” and the values of learning is the best way to serve as an official, with fair competition and equal selection of the best as its fundamental principles. In the practice of imperial examinations in 1300, it has cultural phenomena such as art, history, and folklore that are integrated, and a wealth of imperial examination documents and other related material cultural relics have been accumulated.

3. Method

Our proposed approach is based on deep belief networks (DBN), which can be seen as a stacked combination of many RBMs. The feature learning of the model occurs in the pretraining stage, while its weights and biases are adjusted under the guidance of a trainer in the fine-tuning stage. This section discusses DBN-based model, its training process, and the parameter determination in detail.

3.1. DBN-Based Short-Term Traffic Flow Prediction Model

3.1.1. Restricted Boltzmann Machines

DBN can be regarded as a stack of restricted Boltzmann machines (RBM) in structure. RBM is a two-layer undirected graph consisting of the underlying visible layer and the upper hidden layer. Its structure is shown in Figure 1. Slightly different from the classic Boltzmann machine (BM), RBM only has connections between layers, and no connections between neurons within a layer. This reduces the size of the neural network and simplifies the training process, which is beneficial for practical applications. RBM is faster than BM due to restrictions in terms of connections between nodes. It is expressive to encode any distribution, computationally efficient and improved performance [31, 32].

Where v denotes the set of neurons in the visible layer in RBM, the set of neurons in the hidden layer is denoted by h, while v_i and h_j are the neurons in the visible layer and the hidden layer, respectively. m and n are the number of neurons in the visible layer and the hidden layer, respectively. It is assumed that each neuron in the RBM satisfies the binomial distribution, that is, ∈{0, 1} and h_j∈{0, 1}. Each neuron has an activation function σ(x), usually the sigmoid function is selected as the activation function, and its mathematical form is

When a neuron state is set to 1, the neuron is activated. RBM is an energy model whose energy function can be expressed aswhere is the weight between the ith neuron in the display layer and the jth neuron in the hidden layer, is the bias of the ith neuron in the display layer, and is the bias of the jth neuron in the hidden layer. , , and are the parameters that RBM needs to learn, and these three parameters are represented as , that is, .

According to the principles of statistical mechanics, the neuron state (v, h) has the following joint probability distribution function:where is a normalization factor, also known as the partition function.

Similarly, other joint probability and conditional probability distribution functions can also be expressed in this way:

The joint probability distribution is represented by the energy equation of RBM, and then its conditional probability distribution function is obtained, which lays the foundation for the description of the training algorithm of RBM.

3.1.2. RBM Training Process

The training goal of RBM is to make its structural parameters best fit the training data, and this process is accomplished by adjusting the parameter θ of the RBM. In practical applications, for the determination of parameters, the method of maximum likelihood estimation is usually used, which is realized by finding the maximum value of the likelihood function. The constructed likelihood function is as follows:where is the number of training samples in the training data set and is the state of the neurons in the display layer corresponding to the sth training sample, that is, the state of the neurons in the display layer when the sth training sample is input.

The partial derivative of the likelihood function with respect to the parameters can be obtained as its gradient value:

Then the gradient update process of the tth step can be expressed aswhere is the learning rate.

Generally, in order to prevent the training process from falling into a local minimum and to solve the contradiction between the training speed and the convergence of the training process, the momentum method is usually used in the training process, that is, a momentum factor is added when updating the gradient. Therefore, (6) can be transformed into the following form:where is the momentum factor.

The above is the basic process of RBM training. However, when calculating the gradient, it is found that the second item in parentheses is difficult to calculate directly. In the iterative process, each calculation of this item needs to traverse all the states of the explicit layer and the hidden layer, which is undoubtedly catastrophic for the calculation process. For functions that are difficult to calculate directly, we can use the sampled value approximation instead, which uses the Gibbs sampling method. The sampled value is obtained by sampling the objective function, and the sampled value is used instead of the expected value. If the standard Gibbs sampling method is used to sample the objective function, after each sample sequence is obtained, the preheating part in the sequence needs to be removed. That is to say, before the Markov chain reaches the steady-state distribution, it will take some time to prepare, which will affect the system efficiency in practical applications. To solve this problem, a k-step contrastive divergence (CD-k) algorithm can be used. The CD-k algorithm can be regarded as an improvement to the Gibbs sampling method. In the Gibbs sampling algorithm, the initial value is set randomly, so that for the Markov chain, a finite number of state transitions are required to reach the steady-state distribution. In the CD-k algorithm, the input sample value is directly used as the initial value. Since the initial value obeys the target distribution, the Markov chain converges at the beginning, that is, the stable distribution is achieved, which saves a lot of time. After that, the desired sample value can be obtained after only k steps of sampling. Through the sampling process, the expected value is replaced, so that the parameter value of the RBM can be updated.

The training process of RBM can be regarded as the process of reconstructing and fitting the input data by the explicit layer. The effect of RBM training can be evaluated by the reconstruction error. The reconstruction error can be described as follows:where is the number of input samples and is the number of neurons in the display layer, that is, the dimension of the input data. is the input value of the kth neuron in the display layer corresponding to the sth input sample and represents its reconstructed value.

3.1.3. Deep Belief Network and Training Process

The DBN is a typical nonconvolutional network that may be seen as a stacked combination of many RBMs. There are r RBMs that make up the DBN, which are numbered RBM1 to RBMr. Alternatively, the DBN may be thought of as having an input layer and a hidden r-layer. Hidden layers derive data characteristics from input sample data, while the input layer accepts data samples. In comparison to a shallow network, DBN is superior at identifying and extracting data characteristics. From the bottom to the top, the DBN extracts data characteristics. In order to get the most representative data features, the lower RBM's output is sent into the higher RBM and the features it extracts are fed back into the lower RBM. The depth of DBN determines its feature extraction ability, but the number of layers of RBM depends on the specific situation. For data sets with a small amount of data, too many layers may cause overfitting [33–36].

During the DBN training process, there are two stages: the pretraining stage and the fine-tuning stage, respectively. Pretraining is to train the RBM layer by layer from bottom to top, so that each layer of RBM converges to a local minimum during training. All network weights are fine-tuned at this step, so that the model's training process may converge to its global minimum. BP method is employed for this purpose. Pretraining and fine-tuning are performed independently, and the fine-tuning stage needs to use the results of the pretraining stage. For deep neural networks, the training method of pretraining and fine-tuning can achieve a good training effect, and pretraining can solve the problems of gradient disappearance and local minima to a certain extent.(1)Pretraining stage: The pretraining stage is an important stage in the whole training process, and feature learning mainly occurs in this stage. The goal of the pretraining phase is to tune the model parameters to complete the reconstruction of the data. The pretraining process can be decomposed into training the RBM. When the training of RBM1 is completed, its output value is used as the input value of RBM2 to complete the initialization of RBM2. After that, the output value of RBM2 is used as the input value of RBM3 after the training is completed. And so on until the RBMr training is complete. Pretraining is an unsupervised training method that does not need to add labels when constructing training samples.(2)Fine-tuning stage: During the fine-tuning step, the model's weights and biases are adjusted under the guidance of a trainer in order to get parameters that are as close as possible to the global optimal. The BP method is employed extensively in the fine-tuning step, and an output layer generates the output result. The model structure consists of r + 2 layers of neurons, including an input layer, an r hidden layer and an output layer.

The data set in the fine-tuning stage is denoted as , s = 1, 2,…, S, then the error function is in the form:where is the number of neurons in the output layer L^r+2, and and are the real value and output value of the kth neuron corresponding to the sth sample, respectively.

According to the gradient descent algorithm, the weights and biases are updated according to the following process:where is the weight of the jth neuron in the L^u layer and the ith neuron in the L^u−1 layer at the t step, where 2 ≤ u ≤ r + 2. is the bias of the jth neuron in the L^u layer at step t. and are the momentum factor and learning rate of the fine-tuning stage, respectively. is the output value of the ith neuron in the L^u−1 layer corresponding to the sample .

3.2. Model Parameter Determination and Performance Optimization

3.2.1. Parameter Determination

The quality of the training data and the accuracy of the predictions are directly linked to the prediction model's parameter selections during training. The training parameters include the number of hidden layers, the number of input layer neurons, and the number of hidden layer neurons in each layer, as well as the CD-k algorithm and the BP algorithm, which are used in conjunction with each other. In both the pretraining and fine-tuning stages, parameters, such as the learning rate, momentum factor, and so on, must be established. For the adjustment of model parameters, there is currently a lack of theoretical guidance. In practice, the trial-and-error method is usually used in combination with the experience of the parameter adjusters, that is, first set an initial value for all parameters according to experience, and then adjust each parameter one by one. When adjusting a certain parameter, other parameters should be fixed. When using the trial and error method to adjust parameters, attention should be paid to the order of parameter adjustment, and the parameters related to the model structure can be adjusted first.

3.2.2. Performance Optimization

(1)Local minima problem: deep networks often encounter local minima problems when performing gradient descent. The training process may oscillate at the local minima, increasing training time and even affecting convergence. For the local minimum problem, the additional momentum method is generally used to solve it, which has been mentioned in the previous description. In this paper, the momentum method is added to the training process of DBN. Adding the gradient update amount of the previous time point in the weight update process can effectively avoid the training process from falling into a local minimum.(2)Gradient disappearance problem: when using the BP algorithm to train a deep network, in the process of error back propagation, the gradient value of the bottom layer of the network is relatively small, and the update speed of weights and biases is slower than that of the upper layer, resulting in insufficient training of the bottom layer, this phenomenon is called gradient vanishing. In the process of gradient descent, using the sigmoid function as the activation function is prone to the problem of gradient disappearance. When the network performs gradient update, the derivative needs to be multiplied continuously. Therefore, the gradient value of the network from top to bottom will become smaller and smaller, resulting in the problem of gradient disappearance. The training mode of pretraining and fine-tuning can alleviate the gradient vanishing problem to a certain extent. But to fundamentally solve it, the ReLU function needs to be used. It can be seen from the expression of the ReLU function that when the value of the independent variable is positive, its derivative value is always 1. Therefore, if the ReLU function is used instead of the sigmoid function as the activation function, the problem of gradient disappearance can be fundamentally solved. In the training process of DBN, the ReLU function is used as the activation function in this paper.(3)Overfitting problem: the solutions to the overfitting problem include regularization, dropout, data set augmentation, etc. Through comparative analysis, this paper chooses the dropout method to eliminate the overfitting phenomenon. The specific experimental results are given in Chapter 4.

3.3. The Evaluation Index of the Modern Value of Imperial Examination Culture

The basic principle of modern value evaluation of imperial examination culture is a unique creative activity. Therefore, the evaluation of such innovative behaviors with highly overlapping knowledge and information must be carefully discriminated and determined based on scientific and reasonable principles to ensure that the imagination and creativity of the imperial examination culture are not damaged by the negative effects of evaluation.(1)The principle of value neutrality: the most frequently used evaluation methods for cross-science or interdisciplinary research in the world are peer review and bibliometric methods, which are also two evaluation models with relatively neutral value. Due to factors such as the ability and subjectivity of reviewing experts, competition brought about by the integration of production, education and research, and scoring and voting to win more, peer review may also damage the fairness and effectiveness of the review, forming a “Matthew effect” or a network of acquaintances, and the more senior experts are, the more likely they are to disagree on emerging cross-cutting or interdisciplinary research. Bibliometrics has its own limitations, such as the incomparability of different disciplines, the imbalance of scientific research capabilities in different countries, and the limitations of emerging or small disciplines and their researchers. Therefore, its objectivity is relative. In order to maintain value neutrality, the quantitative methods of mature disciplines cannot be used to evaluate unpopular disciplines such as imperial examination culture.(2)The principle of efficacy delay: academic research, like talent training, has a periodic and lag effect. Evaluating the quality or effectiveness of a research paradigm or a research methodology should not be limited to whether it is ancient or modern, Chinese or foreign, classic or new. Instead, it should be seen whether it is suitable for the research object, and whether its application effectively promotes the deepening or expansion of the academic circle's understanding of the research object. This principle is also applicable to the effect judgment of the interdisciplinary research methodology applied to the modern value of imperial examination culture. For the evaluation of the modern value of the imperial examination culture, various direct or indirect ways can be considered to improve the effectiveness of the evaluation standards.(3)Sustainable dialectical principle: just like the evolution of natural and social phenomena, the modern value of imperial examination culture also has an evolutionary process. The study of the modern value of imperial examination culture is not a static structure, but a dynamic extension, a continuous existence that may be unpredictable and nonrepetitive. Its development is a living, life-rich creative and evolutionary process, which is not only stably connected with the history of imperial examinations but also organically connected with modern examinations. It is a continuous evolution and comprehensive integration of many elements. Therefore, we should evaluate the modern value of imperial examination culture from a developmental and dialectical perspective. Based on the above principles, this paper constructs an evaluation index system for the modern value of imperial examination culture, as shown in Table 1.

4. Experiments and Analysis

4.1. Data Samples

We constructed a data set based on the modern value evaluation index system of imperial examination culture in Section 3. The data set includes 1600 sets of data, of which 1440 sets are used as training sets and the rest are used as test sets.

4.2. Performance Optimization Test

In order to solve the overfitting problem, this paper chooses the dropout method to eliminate the overfitting phenomenon. Figure 2 is a comparison chart of the training effect of the DBN model before and after adding dropout. In this paper, the dropout ratio is set to 0.1, that is, 10% of neurons are randomly dropped in each layer of the network. Before adding dropout, after about 300 iterations, the model overfitted and the training error were smaller than the test error. After adding dropout, the training error is about the same as the test error after about 200 iterations. It can be seen that the dropout method can effectively solve the problem of overfitting, so as to obtain a model with stronger generalization ability.

(a)

(b)

4.3. Model Parameter Determination

The determination of model parameters currently lacks the guidance of systematic theory, and the combination of empirical method and trial-and-error method is generally adopted. First set the initial value of the model according to experience, then fix other parameters unchanged, manually adjust a parameter, and so on, until all parameters are adjusted, so that the training process converges smoothly. The number of neurons in the output layer of the DBN model is set to 1, which is the evaluation value of the modern value of the imperial examination culture. The results obtained are shown in Figure 3. It can be seen that when the number of neurons in the hidden layer is 8, the MSE is the smallest. Therefore, the number of neurons in the hidden layer is chosen to be 8.

The learning rate includes the learning rate λ_p in the pretraining stage and the learning rate λ_f in the fine-tuning stage, which are adjusted separately. In the pretraining stage, a λ_p is shared for all RBMs, and its parameter adjustment range is [0.2, 1.0]. The parameter adjustment range of λ_f is [0.2, 3], and the adjustment step size of both is 0.1. The effect of the learning rate on the error is shown in Figure 4. As λ_p increases, the error gradually decreases. When λ_p is greater than 0.8, the error increases slightly, so λ_p is set to 0.8. The variation trend of the error with the value of λ_f is similar to that of λ_p, and the final value of λ_f is set to 2.5. The adjustment process of the momentum factor is similar.

(a)

(b)

The number of iterations should be set not only to satisfy the conditions of stable convergence of the system but also to make the number of iterations as small as possible to reduce the training time. The effect of the number of iterations in the fine-tuning stage on the error is shown in Figure 5. The error gradually decreases as the number of iterations increases, and the final number of iterations is set to 1000. The same method can be used for the pretraining stage, where the number of iterations is finally set to 200.

4.4. Comparison of Model Evaluation Results

In order to prove the accuracy of the DBN model proposed in this paper on the evaluation of the modern value of imperial examination culture, this paper compares the output results of the model with the evaluation results of experts, as shown in Table 2. It can be seen that the output of the model is very close to the evaluation results of experts. In many of our experiments, the results are the same as those obtained from the expert evaluation, or very close to them. In a few experiments, we noticed a small difference of up to 0.4%. The error is small, and the model has superior performance in evaluating the modern value of imperial examination culture.

5. Conclusion

The history of time is surging, and the fair selection of talents will last forever. The culture of “harmony” for more than 2,000 years and the culture of the imperial examination system for more than a thousand years have made the examination culture a Chinese gene. China’s existing examination system is not only the most equitable student selection and educational resource allocation system at this stage but also has an extremely profound and extensive cultural foundation for imperial examinations. Imperial examination culture and examination culture are the soul of talent selection systems such as ancient and modern Chinese and foreign examinations. Cultural treasures representing Chinese tradition must be safeguarded and managed, while study into and use of these resources must be strengthened. This paper conducts an in-depth study on the modern value of the imperial examination culture in the context of cultural confidence and then constructs appropriate evaluation indicators to use the deep learning model to evaluate the value. We introduce the research progress of imperial examination culture interdisciplinary and modern utilization value of imperial examination culture and introduce the technical principle and performance optimization method of DBN neural network. We construct the modern value evaluation system of imperial examination culture and optimally select the parameters of the DBN model through experiments. Compared with the evaluation results of experts, the DBN model proposed in this paper has superior performance in the evaluation of the modern value of imperial examination culture and the error rate is low. The data set we used is small scale, and a large-scale data set will give more insights about the performance of the model. We intend to extend our work by using a large-scale data set, big data analytics tools and different other deep learning algorithms to be able to handle large volumes of data.

Data Availability

The data sets used during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

Thank Ruohan Ma (University College London, SSEES) for her help and support.

References

C. Wang, “Interpretation of xi jinping's exposition on the spiritual independence of the Chinese nation,” Int'l J. Soc. Sci. Stud, vol. 9, p. 31, 2021.
View at: Google Scholar
H. Gan, “Chinese education tradition-the imperial examination system in feudal China,” Journal of Management and Social Sciences, vol. 4, no. 2, pp. 115–133, 2008.
View at: Google Scholar
Y. Bai, “Farewell to confucianism: the modernizing effect of dismantling China's imperial examination system,” Journal of Development Economics, vol. 141, Article ID 102382, 2019.
View at: Publisher Site | Google Scholar
Y. Feng, “From the imperial examination to the national college entrance examination: the dynamics of political centralism in China's educational enterprise,” Journal of Contemporary China, vol. 4, no. 8, pp. 28–56, 1995.
View at: Publisher Site | Google Scholar
K. H. Ko, “A brief history of imperial examination and its influences,” Society, vol. 54, no. 3, pp. 272–278, 2017.
View at: Publisher Site | Google Scholar
X. Kong, X. Zhang, C. Yan, and K. C. Ho, “China's historical imperial examination system and corporate social responsibility,” Pacific-Basin Finance Journal, vol. 72, Article ID 101734, 2022.
View at: Publisher Site | Google Scholar
L. Haifeng, “Rehabilitation of the imperial examination system,” Frontiers of Education in China, vol. 1, no. 2, pp. 300–315, 2006.
View at: Google Scholar
S. Chu, A. Avery, J. Yoshimoto, and J. N. Bryan, “Genome wide exploration of the methylome in aggressive B-cell lymphoma in Golden Retrievers reveals a conserved hypermethylome,” Epigenetics, vol. 2022, Article ID 2105033, 17 pages, 2022.
View at: Publisher Site | Google Scholar
M. Li, H. Liu, and W. C. Huang, “A historical investigation and classification analysis of seven controversies on the reform and abolition of the 1300-year Chinese imperial examination system,” International Journal of Educational Research, vol. 114, Article ID 101976, 2022.
View at: Publisher Site | Google Scholar
P. Williams, “Equity in the imperial examination (keju) and its influences on the social mobility of ancient rural China,” Enrollment & Examination in Hubei, vol. 54, no. 4, pp. 511–523, 2007.
View at: Google Scholar
Y. L. Dong, “Analysis on the different circumstances of the modern imperial examination system in west and China,” Academics, vol. 1, 2016.
View at: Google Scholar
W. L. Peng, “A brief analysis on the relationship between imperial examination system and ancient feudal politics,” Journal of Jishou University (Natural Sciences Edition), vol. 30, no. 1, p. 34, 2009.
View at: Google Scholar
W. Yu, “The examination culture in imperial China,” Asia Pacific Journal of Education, vol. 39, no. 2, pp. 265-266, 2019.
View at: Publisher Site | Google Scholar
G. Q. Yang, “The restriction of the imperial examination system caused by the turmoil in the late qing dynasty,” Studies in Qing History, no. 2, p. 1, 2021.
View at: Google Scholar
Z. G. Wang and S. H. Zang, “Selection of medical talents under the influence of imperial examination system in the Tang Dynasty,” Zhonghua yi shi za zhi, vol. 35, no. 4, pp. 198–201, 2005.
View at: Google Scholar
Z. Yun, “Fan wencheng and the transplantation of the ming dynasty imperial examination system in the early qing dynasty,” Journal of East China Normal University, vol. 46, no. 1, p. 85, 2014.
View at: Google Scholar
Y. B. Zhuge, “The discussion of farewell Articles in Song dynasty s imperial examination,” Journal of Renmin University of China, vol. 33, no. 4, p. 125, 2019.
View at: Google Scholar
W. L. Peng and Y. H. Su, “Practice and influence of the ancient imperial examination system in the minority-concentrated regions,” Journal of Jishou University (Natural Sciences Edition), vol. 33, no. 2, p. 41, 2012.
View at: Google Scholar
Y. Liu, “The historical reasons and institutional significances of the poetry component of the imperial examination in the qing dynasty,” Studies in Qing History, vol. 2021, no. 2, p. 90, 2021.
View at: Google Scholar
G. Yang, M. Mozzicafreddo, P. Ballarini, S. Pucciarelli, and C. Miceli, “An in-silico comparative study of lipases from the antarctic psychrophilic ciliate Euplotes focardii and the mesophilic congeneric species Euplotes crassus: insight into molecular cold-adaptation,” Marine Drugs, vol. 19, no. 2, p. 67, 2021.
View at: Publisher Site | Google Scholar
C. H. Du, “Probe profound truth of imperial examination system of dali kingdom,” Journal of Dali University, vol. 10, no. 7, pp. 1–5, 2011.
View at: Google Scholar
J. G. Zhang, “A study on the imperial examination from the five Dynasties to the early Song dynasty,” Journal of Xinyang Normal University, vol. 38, no. 3, pp. 129–133, 2018.
View at: Google Scholar
J. Wang and Y. Sun, “A textual research and explanation of the imperial examination vocabulary” bright number” and” dark number,” International Journal of Social Science and Education Research, vol. 4, no. 4, pp. 455–458, 2021.
View at: Google Scholar
K. Li, Y. Qian, D. Xu, J. Zeng, M. Wang, and J. Zhang, “On the trace of scatter matrix difference: a convex approximate discriminant analysis formulation and its mechanism research in imperial examination system,” The International Journal on Artificial Intelligence Tools, vol. 24, no. 05, Article ID 1550015, 2015.
View at: Publisher Site | Google Scholar
Q. Jiang and L. Z. X. lineages, “Gentry and imperial examination foundations in songyang county, zhejiang province in late qing China,” Studies in Qing History, vol. 122, no. 6, pp. 98–114, 2020.
View at: Google Scholar
H. J. Zhang, “The critical reflection on the imperial examination and its characteristics,” Journal of East China Normal University, vol. 33, no. 4, p. 107, 2015.
View at: Google Scholar
X. Z. Ding, “Taking advantage of the imperial examination: on disproportionate reading of the classics in the imperial examination era-also on the geographical distribution of successful candidates in the examination,” Journal of East China Normal University, vol. 33, no. 4, p. 99, 2015.
View at: Google Scholar
Y. Q. Li and H. Y. Ma, “Influence of Chinese imperial examination on the western civil service system,” Journal of Renmin University of China, vol. 22, no. 1, pp. 112–117, 2008.
View at: Google Scholar
B. A. Elman, “The civil examination system in late Imperial China, 1400–1900,” Frontiers of History in China, vol. 8, no. 1, pp. 32–50, 2013.
View at: Google Scholar
C. Chen, H. Li, H. Li, Y. Liu, and S. Wan, “Efficiency and Fairness Oriented Dynamic Task Offloading in Internet of Vehicles,” IEEE Transactions on Green Communications and Networking, vol. 6, 2022.
View at: Google Scholar
N. Zhang, S. Ding, J. Zhang, and Y. Xue, “An overview on restricted Boltzmann machines,” Neurocomputing, vol. 275, pp. 1186–1199, 2018.
View at: Publisher Site | Google Scholar
H. Larochelle, M. Mandel, R. Pascanu, and Y. Bengio, “Learning algorithms for the classification restricted Boltzmann machine,” Journal of Machine Learning Research, vol. 13, pp. 643–669, 2012.
View at: Google Scholar
R. Ali, A. M. Khatak, F. Chow, and S. Lee, “A case-based meta-learning and reasoning framework for classifiers selection,” in Proceedings of the 12th international conference on ubiquitous information management and communication, pp. 1–6, ACM, New York, NY, USA, January 2018.
View at: Google Scholar
R. Ali, S. Lee, and T. C. Chung, “Accurate multi-criteria decision making methodology for recommending machine learning algorithm,” Expert Systems with Applications, vol. 71, pp. 257–278, 2017.
View at: Publisher Site | Google Scholar
Y. Hua, J. Guo, and H. Zhao, “Deep belief networks and deep learning,” in Proceedings of the 2015International Conference on Intelligent Computing and Internet of Things, pp. 1–4, IEEE, Harbin, January, 2015.
View at: Google Scholar
Y. Wei and Z. Weng, “Research on TE process fault diagnosis method based on DBN and dropout,” Canadian Journal of Chemical Engineering, vol. 98, no. 6, pp. 1293–1306, 2020.
View at: Publisher Site | Google Scholar

Copyright

Copyright © 2022 Xiu Zhao. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Mobile Information Systems

Federated Learning, Internet of Things, and Edge Computing for Smart Services

[Retracted] Evaluation of the Modern Value of Imperial Examination Culture in the Context of Cultural Confidence Based on Deep Learning Models

Abstract

1. Introduction

2. Related Work

3. Method

3.1. DBN-Based Short-Term Traffic Flow Prediction Model

3.1.1. Restricted Boltzmann Machines

3.1.2. RBM Training Process

3.1.3. Deep Belief Network and Training Process

3.2. Model Parameter Determination and Performance Optimization

3.2.1. Parameter Determination

3.2.2. Performance Optimization

3.3. The Evaluation Index of the Modern Value of Imperial Examination Culture

4. Experiments and Analysis

4.1. Data Samples

4.2. Performance Optimization Test

4.3. Model Parameter Determination

4.4. Comparison of Model Evaluation Results

5. Conclusion

Data Availability

Conflicts of Interest

Acknowledgments

References

Copyright