Improving the effectiveness of traditional education based on computer artificial intelligence and neural network system

Abstract

The application of artificial intelligence and machine learning algorithms in education reform is an inevitable trend of teaching development. In order to improve the teaching intelligence, this paper builds an auxiliary teaching system based on computer artificial intelligence and neural network based on the traditional teaching model. Moreover, in this paper, the optimization strategy is adopted in the TLBO algorithm to reduce the running time of the algorithm, and the extracurricular learning mechanism is introduced to increase the adjustable parameters, which is conducive to the algorithm jumping out of the local optimum. In addition, in this paper, the crowding factor in the fish school algorithm is used to define the degree or restraint of teachers’ control over students. At the same time, students in the crowded range gather near the teacher, and some students who are difficult to restrain perform the following behavior to follow the top students. Finally, this study builds a model based on actual needs, and designs a control experiment to verify the system performance. The results show that the system constructed in this paper has good performance and can provide a theoretical reference for related research.

Keywords

Computer artificial intelligence neural network education

1 Introduction

The more prosperous education, the more prosperous the country, and the stronger the education, the stronger the country. Education is the basis for a country to become rich and powerful. Our country is now in a transition period from an economic power to a world power. Therefore, higher education is needed to provide more medium- and high-end technical and skilled talents to the country. In particular, the “New People Promotion Law” implemented in 2017 loosened the development of colleges and universities so that they can quickly respond to the needs of society and the country and train qualified personnel. As artificial intelligence is promoted to the national strategy, it will have a profound impact on various industries in my country. Universities should keep up with the trend of the times, grasp the footsteps of the “artificial intelligence” era, and apply “artificial intelligence” to the education and training of students. However, how to adopt “artificial intelligence +” is still a problem. This article will discuss how the application of artificial intelligence to teaching management in colleges and universities will affect the development of the school [1]. If colleges and universities want to better integrate artificial intelligence technology with teaching management, they need to have a deep understanding of artificial intelligence technology and have a clear direction and overall design on how to use it. Moreover, this study explores what changes and how those technologies of artificial intelligence can be applied to teaching management. Moreover, this study discusses the application countermeasures of intelligent applications in teaching, which has guiding significance for teachers to use intelligent optimization teaching in the future, students to use intelligent applications to achieve personalized learning, and teaching managers to use intelligent applications to improve work efficiency. It has become a reality for universities to apply artificial intelligence technology to teaching and learning. Moreover, teachers use artificial intelligence to assist themselves in the process analysis of pre-class preparation, teaching courses, correcting homework and after-school counseling. Through the study of the process of artificial intelligence on students’ personalized learning, autonomous learning, and after-school practice, it is found that artificial intelligence can change the disadvantages of traditional classroom teaching and make teaching and learning more intelligent, precise, and personalized [2]. The use of artificial intelligence can make teaching management more efficient, simplify the process of teaching management, reduce the cost of teaching management, and improve the level of teaching management. The use of artificial intelligence technology in teaching management can make the management more just and transparent, make the rights and responsibilities clear, and make managers more responsible. The intelligent management system can make the management data public, and can enable teachers, students, parents, etc. to participate in the teaching management, so that the teaching management is supervised by everyone and do a good job in service. What an enterprise sells are a product, and whether this product is popular in the market is a criterion that determines the quality of this enterprise. What the school “sells” are excellent “talents” that meet the development requirements of the enterprise, can bring performance to the enterprise, and make the enterprise healthy and sustainable. Employment is the foundation of national development and one of the important indicators of national economic development to measure the stability of the country’s political constitution. Therefore, whether the students trained by colleges and universities can meet the needs of enterprises after graduation and whether the training results can meet the requirements of enterprises in their positions is an important issue that all colleges and universities in China must face. This requires colleges and universities to work hard on talent training and teaching management optimization, and should closely follow the development of the times, seize new technologies and new opportunities, and apply new technologies such as artificial intelligence, Internet of Things, and information technology to teaching services. Moreover, colleges and universities need to intelligently manage matters such as teaching management models, teacher teaching, student learning, comprehensive evaluation, and student employment evaluation, and improve efficiency to train more outstanding talents for the school and build their own brands [3].

2 Related work

The literature [4] made a breakthrough in the field of deep learning of neural networks, so that humans once again saw the possibility of machines achieving intelligence and surpassing humans, which is a landmark event in the progress of artificial intelligence technology. The literature [5] in the field of Go, where humans believe that machines cannot beat humans, defeated South Korean chess player Li Shishi, which once again triggered an upsurge in artificial intelligence. As Internet giants such as Google, Baidu, Ali, and Tencent entered the battlefield of artificial intelligence research and development, another round of artificial intelligence frenzy was launched. At present, artificial intelligence technology is increasingly mature and applied to people’s daily life. People are happy and need them to integrate into life. This wave of artificial intelligence may realize the intelligence of human beings and make science fiction movies a reality. Artificial intelligence is the scientific technology that studies the way people think and learn through machines, and is the scientific technology that extends human intelligence, such as: speech recognition, deep learning, sentiment analysis, learning analysis, and intelligent behavior. Artificial intelligence allows machines to think and act like humans, and eventually replace humans for tasks that only humans can accomplish. The literature [6] proposed that computers or robots using artificial intelligence systems can evolve on their own and can produce more advanced artificial intelligence systems than themselves. Faced with artificial intelligence, we can neither overestimate its impact on us nor underestimate the convenience it brings to life and work. The main research technologies of artificial intelligence are expert system [7], natural language understanding [8], machine learning [9], distributed artificial intelligence, machine learning [10], pattern recognition [11] and so on. In recent years, the rapid development of computer hardware technology has led to a significant increase in computer performance. The global popularity of the Internet provides a large amount of data foundation for big data technology and deep learning technology, so that the two technologies have been further developed. Artificial intelligence has achieved rapid development and is widely used in many fields such as family life, medical care, transportation and so on. Moreover, various industries are actively exploring how to use artificial intelligence technology to solve industry problems, and education is no exception [12]. The literature [13] believed that artificial intelligence is a technique for increasing ability, using ability and empowering ability, which is divided into subjectivity and assistance and used in teaching. Subjectivity refers to teaching systems built using artificial intelligence technology, such as intelligent teaching assistants, intelligent teaching platforms, intelligent robots, etc. Assistance refers to integrating functional modules or some technologies of artificial intelligence into teaching resources, teaching tools, teaching management, teaching evaluation, and transforming them into media or tools to exert their functions, such as intelligent teaching platforms and adaptive learning platforms, intelligent teaching management, intelligent evaluation, etc. [14]. Only a combination of artificial intelligence, VR, AR, behavior analysis, learning analysis, and intelligent robot technology can have a profound impact on teaching. The impact of using only one technology on teaching is very limited [15]. Therefore, the research and application of machine learning, natural language understanding, pattern recognition, big data, learning analysis and other technologies to teaching are the opportunities and challenges brought by artificial intelligence to teaching. Pattern recognition is to use a computer to identify existing things, and then put it into the same or similar patterns. By studying how to make the computer recognize objects, videos, audio and other information, the computer can listen to sounds, see objects and feel nature like humans. To apply pattern recognition to teaching, we must first collect students’ voice, emotion, behavior and other data, and carry out in-depth analysis of these data to obtain a data model. According to this data model, the machine will provide students with personalized learning services [16]. Pattern recognition can be used in teaching to improve teaching effectiveness. For example, in the practical operation course, the machine can scan the student’s operation and compare it with the standard operation to correct and guide the student’s operation. A large amount of high-quality data provides the most fundamental support for the realization of artificial intelligence. The traditional data volume is small, the diversity is insufficient, and the circulation speed is not high, while the big data has the characteristics of huge data volume, high-speed circulation, diversity, and authenticity [17]. The in-depth development of big data technology has brought opportunities for machine learning and intelligence as well as opportunities for future intelligent education [18].

3 Teaching optimization algorithm

The specific implementation steps of the TLBO algorithm are as follows [19]:

Step 1: The number of students P_n, the maximum number of iterations G_max, the number of courses D_n, and the upper and lower limits of course grade (U_L, L_L) are initialized. A student j has a grade of $X^{j} = (x_{1}^{j}, x_{2}^{j}, \dots, x_{D_{n}}^{j}), j = 1, 2, \dots, P_{n}$ in each subject.

Step 2: Population initialization, that is, all students and teachers form a class, which is expressed by the following matrix: $Class = [\begin{matrix} x_{1}^{1} & x_{2}^{1} & \dots & x_{D_{n}}^{1} \\ x_{1}^{2} & x_{2}^{2} & \dots & x_{D_{n}}^{2} \\ ⋮ & ⋮ & ⋮ \\ x_{1}^{D_{n}} & x_{2}^{D_{n}} & \dots & x_{D_{n}}^{P_{n}} \end{matrix}]$ (1)

Among them, the column average of the matrix represents the average score M_i of all students in different courses.

Step 3: We assume that the grade distribution of students in the class meets the normal distribution, the distribution function is $f (X) = \frac{1}{σ \sqrt{2 π}} e^{\frac{- (x - μ)}{2 σ^{2}}}$ , σ² is the variance, and μ is the average score. The student with the best grade in the class (the best fitness value) is selected as the teacher X_teacher.

Step 4: The teaching process of the first stage of the algorithm is shown in Fig. 1 above. The students’ grades are relatively low, and their average score is Mean_A. At this time, the best grade T_A is the teacher. After a certain period of teaching, the student’s performance was improved to be close to the teacher’s performance Mean_B. At this time, it was difficult for the teacher T_A to take up the teaching role, and the teacher changed to the best performing T_B at this time. In the first process, the improvement of student performance is mainly achieved by the difference between the teacher’s performance and the student’s average score [20]. that is: $Different_m {ean}_{i} = r_{i} (X_{teacher} - T_{F} \cdot M_{i})$ (2)

Fig. 1

Schematic diagram of teaching optimization algorithm.

Among them, r_i is a random number in the range of [0, 1]. The teaching factor T_F = round [1 + rand (0, 1) { 2 - 1 }] is a random value between 1 and 2.

According to formula (2), the updated formula of the teaching process is obtained: $\begin{matrix} X_{new} = X_{old} + Different_m {ean}_{i} \\ = X_{old} + r_{i} (X_{tescher} - T_{F} \cdot M_{i}) \end{matrix}$ (3)

Step 5: The second stage of the algorithm is the learning process. The two students were arbitrarily chosen to communicate, report, discuss, etc., so that the current students can gain more knowledge from better classmates, thereby improving their grades. The two classmates X_i and X_j were randomly selected. When X_i ’s grade $X_{i}^{'}$ after teaching was better than X_j ’s grade $X_{j}^{'}$ after teaching, the following formula was used to update the current grade to X_new [21]. $X_{new} = X_{old} + r_{i} (X_{i} - X_{j})$ (4)

Otherwise, if X_j ’s grade $X_{j}^{'}$ is better after the teaching process, the grade is updated by using the following formula: $X_{new} = X_{old} + r_{i} (X_{j} - X_{i})$ (5)

The updated score cannot be regarded as the final score, and it can be determined only by comparing fitness values. Only if the fitness value of X_new is better than X_old, X_old is updated. Otherwise the original value is retained.

Step 6: If the termination condition is satisfied, it will end and output the final result. Otherwise step 4 is returned.

The basic TLBO algorithm flow chart is shown in Fig. 2 [22].

Fig. 2

Basic teaching optimization algorithm flow chart.

4 Particle swarm optimization

Each particle in the particle swarm optimization (PSO) algorithm is a vector, which is used as a solution in the optimization problem. Each PSO particle in the D-dimensional space has three attributes: the current position x_i = [x_i,1, x_i,2, ⋯ , x_i,D], the particle historical optimal position p_i = [p_i,1, p_i,2, ⋯ , p_i,D], and the velocity v_i = [v_i,1, v_i,2, ⋯ , v_i,D]. For the function optimization problem, each particle in the particle swarm optimization algorithm determines its next position through its own historical optimal position and its neighborhood optimal or global optimal position, and its own speed adjustment direction and step size. The particle position update method in the PSO algorithm is shown in Fig. 3 [23].

Fig. 3

PSO particle position update map.

Among them, the particle velocity and position update formulas are as follows: $\begin{matrix} v_{i} (t + 1) = ω \cdot v_{i} (t) + c_{1} \cdot (p_{best, i} (t) - x_{i} (t)) \\ + c_{2} \cdot (g_{best, i} (t) - x_{i} (t)) \end{matrix}$ (6)

In the formula, v_i (t + 1) is the speed at the next moment, ω is the speed weight, and v_i (t) is the current speed. Meanwhile, c₁c₂ is the learning factor, x_i (t + 1) is the position at the next moment, an p_best,i (t) is the current optimal position. g_best,i (t) is the optimal position in the neighborhood, and x_i (t) is the current location. In the above formula, the ω value in the first inertial term has a greater influence on the flying speed of the particle. The larger the ω value, the faster the flying speed of the particle, which means that the particle will have a larger step size and faster speed to conduct a global search. If the ω value is smaller, it means that the particles will perform a local search at a smaller particle speed. Formula (6) uses the updated speed formula to update the original position [24 –26].

In the calculation of the D-dimensional space search process, the speed and position range of the particle swarm algorithm are [v_min, v_max] and [x_min, x_max], respectively, and the updating formula force of each particle at time t is: $\begin{matrix} v_{i, j} (t + 1) = ω \cdot v_{i, j} (t) + r_{1} c_{1} \cdot (p_{best, i, j} (t) - x_{i, j} (t)) \\ + r_{2} c_{2} \cdot (g_{best, i, j} (t) - x_{i, j} (t)) \end{matrix}$ (7) $x_{i, j} (t + 1) = x_{i, j} (t) + v_{i, j} (t), j = 1, 2, \dots, D$ (8)

Among them, r₁r₂ is a random number between [01]. The first term on the right side of formula (5) is the inertia term, which means that the particles have the characteristics of maintaining the original speed of movement and are used to balance the ability of global search and local search. The second term is the cognition term, which means that the particles have the characteristic of remembering their own optimal position, so as to enhance the algorithm’s global search ability and avoid falling into the local optimum. The third item is the social item, which has the characteristic of optimal approach to the neighborhood, indicating that the particles communicate and share operations.

The implementation steps of the PSO algorithm are as follows:

Step 1: The population size, the maximum number of iterations, the speed and position of particles in the population are initialized. Each particle contains D variables, and D is the dimension of the search space. The current position is the historical optimal position of the particle, and the neighborhood optimal position is inf

Step 2: The fitness value of the particle is calculated, the historical optimal position of the particle when the fitness value of the particle is optimal is stored as p_best, and the optimal location of the neighborhood of the particle when the fitness value is optimal is stored as g_best.

Step 3: According to equation (7) and equation (8), the particle velocity and position are updated;

Step 4: The particle fitness value is recalculated. If the updated fitness value is smaller after comparison, the updated position is used instead of the historical optimal position p_best of the particle. At the same time, the current historical optimal value of the particle and the optimal fitness value of the neighborhood are compared, and the better individual is reserved as the optimal position of the domain g_best.

Step 5: It is judged whether the maximum number of iterations has been reached. If the maximum number of iterations has been reached, the optimal solution is output. Otherwise, the algorithm iteration number is increased by 1 and the algorithm returns to step 2.

The flow chart of the PSO algorithm is shown in Fig. 4.

Fig. 4

PSO algorithm flowchart.

5 Hybrid particle swarm teaching optimization algorithm (HPTLBO)

This study designed to add tutor self-study parameters to the teaching optimization algorithm to achieve extracurricular learning, details as follows: $\begin{matrix} X_{new} = X_{old} + w_{tatur} (X_{t} - X_{old}) \\ + X_{old} w_{self} (2 \cdot rand () - 1) \end{matrix}$ (9) $w_{tatur} = w_{max} - G (w_{max} - w_{min}) / G_{max}$ (10)

Among them, w_self is the self-learning weight, X_t is the tutoring score, and rand () is the random number between (0, 1). At the same time, w_tatur is the tutoring weight, G is the current iteration number, and G_max is the maximum number of iterations. w_max is the maximum weight, and w_min is the minimum weight.

It can be seen from the weight formula that as the search process progresses, the impact of tutoring will be correspondingly weakened, and the self-study part may produce a benign effect and may also cause a decline in grades.

The self-optimal and neighborhood optimal trends are balanced by improving the learning factor c₁c₂ in the basic PSO algorithm. At present, there are many ways to choose the learning factor: constant value, linear decrease, c₁ linear decrease, c₂ linear increase, and nonlinear change.

In the intersection area, 10% of the particles are randomly selected as individual particles to be crossed. After the individuals to be crossed are randomly paired with each other, the paired parents are used to generate the position and velocity of the children. $x_{new} = r \cdot x_{p, 1} + (1 - r) \cdot x_{p, 2}$ (11) $v_{new} = | v_{p, 1} | \cdot \frac{v_{p, 1} + v_{p, 2}}{| v_{p, 1} + v_{p, 2} |}$ (12)

Among them, r is the random number between (0, 1), x_p,1 and x_p,2 are the position of the parent individual, and v_p,1 and v_p,2 are the speed of the paired individual. Meanwhile, x_new and v_new are the position and velocity of the individual offspring.

After using replacement operators to replace the particles to be crossed with new particles, a new population is obtained. The new population inherits the original population according to the fitness value.

S-shaped weights are used in the formula (5) instead of fixed weights as the product of speed. The non-linear weight has a small change in the early and late speeds of the iteration, that is, the early search speed is kept at a relatively fast level, and the speed does not drop too fast at the end of the search. When the search speed is in the middle of the S shape, the speed drops faster, which allows the particles to move faster to the global best. The following formula is used to set the weight ω_PSO value: $ω_{PSO} = \frac{ω_{max} - ω_{min}}{e^{(c_{speed} \cdot G / G_{max} - c_{speed})} + 1} + ω_{min}$ (13)

In the formula, ω_max is the maximum weight, ω_min is the minimum weight, and G is the current number of iterations. Meanwhile, G_max is the maximum number of iterations, and c_speed is the speed change adjustment coefficient. As the value of inertia c_speed becomes larger, the falling speed of weight ω_PSO in the middle section gradually becomes larger, that is, the particles in this process will quickly move to the global best.

HPTLBO algorithm implementation steps are described as follows:

Step 1: The population number p_n, dimension D, maximum iteration number G_max, maximum and minimum weights w_min and w_max, external learning parameters w_tatur and w_self, and learning factor c₁c₂ are initialized.

Step 2: Individuals p_n/2 are randomly selected as PartA to execute the optimal extracurricular learning teaching optimization algorithm. The other half as PartB executes the cross-variable weight particle swarm algorithm;

Step 3: The fitness values are calculated separately. In Part A, the selection of teachers is determined, and the teaching/learning process is selected to determine whether to update the student’s grades. In PartB, the historical optimal position of the individual particles and the optimal position of the neighborhood are calculated;

Step 4: PartA performs the self-learning process, PartB updates p_best and g_best, and the fitness value is recalculated

Step 5: The results of PartA and PartB are mixed and with p_mnter probability, mutation is carried out;

Step 6: It is determined whether the maximum number of iterations is reached. If the maximum iteration is reached, the final result is output, otherwise step 2 is returned.

The HPTLBO algorithm flow chart is shown in Fig. 5.

Fig. 5

HPTLBO algorithm flow chart.

6 Teaching Optimization Algorithm (MFTLBO) Inspired by Fish School Behavior

Since the teaching process performed in the teaching process depends on two factors: the teacher’s level X_teacher and the student’s acceptance of the teacher’s content T_F, the presence of the teacher is particularly important.

The random teaching factor proposed in this paper follows the random characteristics of the teaching factor in the basic TLBO, and its calculation method is as follows: $T_{F} = μ - σ \cdot randn ()$ (14) $μ = rand () \cdot (μ_{max} - μ_{min}) + μ_{min}$ (15)

Among them, σ is the variance of the teaching factor, randn () is a random number that meets the standard normal distribution, and rand () is the random number between (0, 1). Meanwhile, μ_max is the upper limit of the average teaching factor, and μ_min is the lower limit of the average teaching factor.

The artificial fish school realizes its visual process through the following methods:

As shown in Fig. 6, the visual range of the artificial fish is Visual, the moving step is Step, the position of the artificial fish at the current moment is X, the viewpoint position at a certain moment is X_V. When the food concentration at the viewpoint position X_V is greater than the food concentration at the current position X, then the artificial fish moves from X to X_next with a certain probability, otherwise it will continue to tour other positions in the field of vision. As the number of artificial fish patrols continues to increase, its knowledge of the situation in the field of vision is more specific, which helps the artificial fish to make corresponding judgments and decisions. The specific implementation formula is as follows: $X_{V} = X + Visual \cdot rand ()$ (16) $X_{next} = X + (2 \cdot rand () - 1) \cdot Step \cdot \frac{X_{V} - X}{∥ X_{V} - X ∥}$ (17)

Fig. 6

Visual map of artificial fish.

Among them, the current position X = (x₁, x₂, ⋯ , x_n), the viewpoint position X_V = (x_{v
₁}, x_{v
₂}, ⋯ , x_{v
_n}), and rand () are random numbers between (0, 1). If the current position is a two-dimensional point, it means that (x₁, x₂) is the center, and any position (y₁, y₂) with the radius of Visual can be visually observed.

There are four basic behaviors of fish schools: foraging behavior, rear-end behavior, group behavior, and random behavior, among them:

(1) Swarm behavior: In order to ensure its own survival advantages or avoid danger, fish groups will instinctively gather to a center. This kind of swarm behavior follows two principles: first, the fish group should be as close as possible to the center of the same type; Second, overcrowding is avoided;

(2) Rear-end behavior: The artificial fish with the highest food concentration in the field of vision is followed, prompting the artificial fish trapped in the local optimum to escape quickly, and speeding up the optimization.

Combined with the description of the above artificial fish school algorithm, in the minimum problem, the rear-end behavior of artificial fish occurs at n_f/δ ≤ 1, and the fish in the field of view all swim towards the optimal fish in the center of the field of view. Otherwise, some of the artificial fish whose food concentration is higher than y_j · n_f/δ choose the rear-end behavior, and the remaining artificial fish choose the foraging behavior. As n_f/δ decreases, the probability that the rear-end behavior is selected is higher. Swarming behavior occurs in a part of mermaids that have a food concentration y_central · n_f/δ higher than the center of the visual field, and the rest of mermaids behave as foraging. The clustering behavior is similar to the rear-end behavior. As n_f/δ decreases, the probability of being selected is higher.

Inspired by rear-end and clustering behaviors in artificial fish schools, this chapter proposes a neighborhood search strategy that uses adaptive Visual, adds a crowding factor δ_l, and specifies that when the number of students surrounding the teacher reaches the upper limit n_upper, that is, when n_upper/δ_l > 1, some students whose grades are better than y_teacher · n_upper/δ_l break away from the teacher’s unified teaching and give up learning to the teacher and perform the learning process. In the teaching process, the fewer the students around the teacher, the better the teaching effect. There are excellent students who account for 5% of the total number of students in the class. As the core person of extracurricular teaching in the learning process, they guide other students and impart knowledge to them. Before making the selection of top students, we should first eliminate the possible same score phenomenon, and ensure that the top 5% of the students’ results are unique, so as to avoid uncertainty in the direction of learning. Similarly, in the process of searching for the best, the fewer students who are supervised by top students, the better the effect.

The steps to implement the neighborhood search proposed in this section are as follows:

Step 1: The parameter crowding factor δ_l and the upper limit of the number of students n_upper is initialized,

Step 2: n_upper/δ_l is calculated to determine whether the teacher is crowded, if n_upper/δ_l ≤ 1, that is, there is still a margin of teaching resources to continue to teach students to perform step 3. Otherwise, some students whose grades are better than y_teacher · n_upper/δ_l will directly perform step 4;

Step 3: The teaching process is executed, and the teachers screened out by the fitness value guide the students to improve the student’s achievements, and enter step 5;

Step 4: A top student in the field of vision and the top 5% of the score is randomly selected as the core to perform clustering behavior, and the level of these students is used as the learning standard for the learning process: $Visual = \frac{1}{m} \sum_{k = 1}^{m} [f (X_{k}^{top}) - f (X_{old})]$ (18) $X_{next} = X_{old} + rand () \cdot \frac{X^{top} - X_{old}}{∥ X^{top} - X_{old} ∥}$ (19)

Among them, Visual is the vision of students. $X_{k}^{top}$ is the top 5% of the top students, m is the total number of top students, $X^{top}$ is a random top student in the field of vision, and the selection basis is $f (X_{k}^{top}) - f (X_{old}) \leq Visual$ .

Step 5: Process effectiveness statistics are carried out. If the failure probability is higher than P_fail, a chaotic search operation is performed.

Validity testing is added to the TLBO algorithm to determine the long-term impact of the teaching process and learning process on the progress of student performance. If the student performance is updated, ac_t and ac_l are separately recorded as a criterion for whether chaotic search is required in the future. The formula for calculating the effectiveness of the teaching process and learning process is: ${ef}_{t} = \frac{{ac}_{t}}{{count}_{t}}$ (20) ${ef}_{l} = \frac{{ac}_{l}}{{count}_{l}}$ (21)

Among them, count_t and count_l are used to count the number of times of entering the teaching process and learning process, respectively. If the ineffectiveness in the teaching and learning process is too large, that is, (1 - ef_t) ≥ P_fail or (1 - ef_l) ≥ P_fail, then the chaotic search operation needs to be performed.

The idea of chaotic sequence optimization algorithm comes from three unique properties of chaotic phenomena: randomness, ergodicity and regularity. In the optimization field, regularity and randomness make chaotic sequences between deterministic and random, with rich spatiotemporal states, and ergodicity can be used as an optimization mechanism to avoid falling into local extremums during the search process.

The description of chaos search is as follows: Chaos is a general non-linear phenomenon, and the motion state with randomness obtained by deterministic equations is generally called chaos. The following sequence can be used to traverse the chaotic sequence of interval [0, 1]: $l_{n + 1} = μ \cdot l_{n} (1 - l_{n})$ (22)

Among them, l_n is the value of the mixed stew variable at the nth iteration, μ is the control parameter of the mixed pure sequence, μ = 4 is taken, and L ={ l₁, l₂, ⋯ , l_m } is the set of chaotic sequences composed of all l_n ∈ (0, 1).

Some students in the teaching process are arranged according to the fitness value. The last 10% of the individuals are adjusted by the chaotic sequence L, and the current teacher’s score is used as the goal to avoid blind self-search: $X_{i}^{new} = X_{t} + L_{i} \cdot (1 - \frac{G}{G_{max} + 1}) \cdot (X_{teacher} - X_{i})$ (23)

Among them, L_i is the i-th number in the L sequence in the chaotic sequence, G is the current number of iterations, and G_max is the maximum number of iterations. Meanwhile, X_teacher is the current teacher.

After the chaotic search, the fitness value of the student is calculated. If the fitness value of the newly acquired student is better, the new score of the student is used to replace the original score. At the same time, from multiple students who have been updated by chaos, according to the fitness value, the new teacher is found to replace the original teacher, and continue to perform the next operation.

7 System performance verification

The educational auxiliary system based on artificial intelligence and neural network constructed in this paper was applied to actual teaching, and the problems in teaching were found, and the teaching process was effectively verified. This article used experimental teaching methods to verify model performance. This study used Class A and Class B for verification, Class A is the experimental group, and Class B is the control group. The experimental group adopted the system constructed in this paper to assist in teaching, found problems and corrected them in time. The experiment lasted for one semester. The results obtained after the experiment are shown in Table 1 and Fig. 7.

Table 1
Statistical table of experimental teaching results

Ranking Class A Class B Ranking Class A Class B

1 92 85 26 69 51

2 91 84 27 67 50

3 91 84 28 67 49

4 91 84 29 66 48

5 89 78 30 63 47

6 88 77 31 62 44

7 87 77 32 62 43

8 84 72 33 59 42

9 83 70 34 59 42

10 82 70 35 59 40

11 81 69 36 58 38

12 77 64 37 58 36

13 77 63 38 58 36

14 76 59 39 55 35

15 75 58 40 55 34

16 75 57 41 54 33

17 74 57 42 54 32

18 74 56 43 53 30

19 72 56 44 53 30

20 72 56 45 53 30

21 72 55 46 52 30

22 71 54 47 51 30

23 70 54 48 49 29

24 70 52 49 47 26

25 70 52 50 47 25

Ranking	Class A	Class B	Ranking	Class A	Class B
1	92	85	26	69	51
2	91	84	27	67	50
3	91	84	28	67	49
4	91	84	29	66	48
5	89	78	30	63	47
6	88	77	31	62	44
7	87	77	32	62	43
8	84	72	33	59	42
9	83	70	34	59	42
10	82	70	35	59	40
11	81	69	36	58	38
12	77	64	37	58	36
13	77	63	38	58	36
14	76	59	39	55	35
15	75	58	40	55	34
16	75	57	41	54	33
17	74	57	42	54	32
18	74	56	43	53	30
19	72	56	44	53	30
20	72	56	45	53	30
21	72	55	46	52	30
22	71	54	47	51	30
23	70	54	48	49	29
24	70	52	49	47	26
25	70	52	50	47	25

Fig. 7

Statistical diagram of experimental teaching results.

As shown in Fig. 7, there is a clear stratification between the experimental group and the control group, that is, the experimental group’s score is significantly higher than the control group. The scores of the two groups of people are subtracted, and the statistical results are shown in Table 2 and Fig. 8.

Table 2

Statistical table of the score difference between the experimental group and the control group

No.	Difference	No.	Difference	No.	Difference
1	7	18	18	35	19
2	7	19	16	36	20
3	8	20	16	37	21
4	8	21	17	38	22
5	11	22	17	39	21
6	11	23	17	40	21
7	11	24	18	41	21
8	12	25	18	42	22
9	13	26	18	43	23
10	12	27	17	44	23
11	12	28	18	45	23
12	14	29	18	46	22
13	13	30	16	47	21
14	17	31	18	48	20
15	17	32	19	49	21
16	18	33	17	50	22
17	17	34	17

Fig. 8

Statistical table of the score difference between the experimental group and the control group.

As shown in Fig. 8, the scores of the experimental group are 7–22 points higher than the control group, which shows that the system constructed in this paper has good performance. When it is applied to teaching, it can effectively improve students’ achievements.

8 Conclusion

Artificial intelligence teaching optimization algorithm is a simple, efficient and fast intelligent optimization algorithm proposed recently. This algorithm has been fully applied in many fields since it was proposed. This paper first briefly discusses the research background, research significance and research status of teaching optimization algorithms, and then explains and analyzes its basic principles, time complexity and space complexity. On this basis, in order to improve the practicability, efficiency and robustness of teaching optimization algorithms, this study proposes a computer-aided teaching system based on computer artificial intelligence and neural networks and a hybrid particle swarm teaching optimization algorithm. The particle swarm algorithm is concise, easy to implement, has fast convergence, and has strong global search capabilities. In addition, this study proposes a teaching optimization algorithm (MFTLBO) inspired by fish behavior. The clustering behavior in the artificial fish swarm algorithm improves the stability of the algorithm convergence, and the following behavior enhances the algorithm search ability. In addition, this study designed control experiments to verify system performance. The results show that the system constructed in this paper has good performance, and when it is applied to teaching, it can effectively improve students’ grades.

Footnotes

Acknowledgment

This paper was supported by (1) The Soft Science Research Project of Hebei Provincial Department of Science and Technology in 2019, “Research on the Cultivation Mode of Teaching Ability of Local Normal University Students Based on The Dual Tutorial System” (No. 18456234); (2) key project of teaching and Research Reform of Cangzhou Normal University in 2019 “Exploration and Practice of Training Mode of “Double Mentors” of Vocational Ability of Normal University Students” (No. 2019JGA002).

References

Wang

Baojian

, Wang

Jing

and Hu

Guoqiang

, College English classroom teaching evaluation based on particle swarm optimization-extreme learning machine model, International Journal of Emerging Technologies in Learning 12(5) (2017), 82–90.

Linhares

R.N.

, Guimarães Alcântara

Caio Mário

, Gonçalves

Everton Ávila

, et al, Teaching evaluation by teachers from Brazil and Portugal: a comparative analysis, American Journal of Educational Research 5(5) (2017), 546–551.

Huang

Ning

, Analysis and design of university teaching evaluation system based on JSP platform, International Journal of Education & Management Engineering 7(3) (2017), 43–50.

Murcia

J.A.M.

, Torregrosa

Y.S.

and Belando

, Questionnaire evaluating teaching competencies in the university environment. Evaluation of teaching competencies in the university, Naer Journal of New Approaches in Educational Research 4(1) (2015), 54–61.

Liu

and Chen

, Research on fuzzy comprehensive evaluation in practice teaching assessment of computer majors, International Journal of Modern Education & Computer Science 7(11) (2015), 12–19.

Zhou

, Li

and Sun

, Teaching performance evaluation by means of a hierarchical multifactorial evaluation model based on type-2 fuzzy sets, Applied Intelligence 46(1) (2016), 1–11.

Lalvarmawi

Florence

, Banik

Uttam

and Anita Devi

, Feedback of medical students on teaching and evaluation methodology in Physiology, National Journal of Physiology Pharmacy & Pharmacology 5(2) (2015), 1–12.

Porozovs

, Liepniece

and Voita

, Evaluation of the teaching methods used in secondary school biology lessons, Nephron Clinical Practice 7(1) (2015), 60–66.

Oliveros

María Amparo

, García

Alejandra

, Valdez

Benjamín

, Evaluation of a teaching sequence regarding science, Technology and Society Values in Higher Education, Creative Education 6(16) (2015), 1768–1775.

10.

Cerón

Manuel Sánchez

, del Sagrario Corte Cruz

Francisca María

, The evaluation of teaching: some consequences for Latin America, Revista Mexicana De Investigacion Educativa 20(67) (2015), 1233–1253.

11.

Heitkamp

S.J.

, Rüttermann

Stefan

, Gerhardt-Szép

Susanne

, Work shadowing in dental teaching practices: evaluation results of a collaborative study between university and general dental practices, Bmc Medical Education 18(1) (2018), 99–104.

12.

, Li

and Ge

, Application of data mining in the colleges’ in-class teaching quality evaluation system, Journal of Computers 10(3) (2015), 166–175.

13.

Angell

and Tewell

, Teaching and un-teaching source evaluation: questioning authority in information literacy instruction, Communications in Information Literacy 11(1) (2017), 95–121.

14.

Brkovic

and Chiles

, Spector – the sustainability inspector’: Participatory teaching, learning and evaluation game for architects, architecture students and pupils, Facta Universitatis 14(1) (2016), 1–20.

15.

Reisenwitz

Timothy H.

, Student evaluation of teaching: an investigation of nonresponse bias in an online context, Journal of Marketing Education 38(4) (2015), 139–144.

16.

Jiang

and Wang

, Evaluation of teaching quality of public physical education in colleges based on the fuzzy evaluation theory, Journal of Computational and Theoretical Nanoence 13(12) (2016), 9848–9851.

17.

Royal

, A guide for making valid interpretations of student evaluation of teaching (SET) Results, Journal of Veterinary Medical Education 44(2) (2016), 1–7.

18.

Eckler

, Greisberger

, Höhne

Franziska

, et al, Blended learning versus traditional teaching-learning-setting: Evaluation of cognitive and affective learning outcomes for the inter-professional field of occupational medicine and prevention / Blended Learning versus traditionelles Lehr-Lernsetting: Evaluierung von kognitiven und affektiven Lernergebnissen für das interprofessionelle Arbeitsfeld Arbeitsmedizin und Prävention, Nephron Clinical Practice 4(2) (2017), 109–121.

19.

Müller

Tjark

, Montano

, Poinstingl

, et al, Evaluation of large-group lectures in medicine – development of the SETMED-L (Student Evaluation of Teaching in MEDical Lectures) questionnaire, Bmc Medical Education 17(1) (2017), 137–142.

20.

Al-Maamari

, Response rate and teaching effectiveness in institutional student evaluation of teaching: a multiple linear regression study, Higher Education Studies 5(6) (2015), 9–15.

21.

Shi

Zhan-Hong

and Zhang

Ding-Hai

, A novel similarity degree of intuitionistic fuzzy sets induced by triangular norm and its application in pattern Recognition, Journal of Intelligent & Fuzzy Systems 37(2) (2019), 3041–3051.

22.

Garofalo

, Mota-Moya

, Munday

, et al, Total extraperitoneal hernia repair: residency teaching program and outcome evaluation, World Journal of Surgery 41(1) (2017), 1–6.

23.

Gong

and Liu

, Consideration of evaluation of teaching at colleges, Open Journal of Social Sciences 04(7) (2016), 82–84.

24.

Yang

Qiang

, Li

Yan-Lai

and Chin

Kwai-Sang

, An ordinal scale-based GDM approach to prioritize customer requirements in QFD product planning, Journal of Intelligent & Fuzzy Systems 37(3) (2019), 4349–4367.

25.

Zhao

Hongzhi

, College physics teaching model design and evaluation research of students’ seriousness, Open Cybernetics & Systemics Journal 9(1) (2015), 2017–2020.

26.

Tran

N.D.

, Reconceptualisation of approaches to teaching evaluation in higher education, Issues in Educational Research 25(1) (2015), 50–61.