Student learning condition prediction model training method and online early warning method based on APO algorithm and neural network

Abstract

The invention discloses a student learning condition prediction model training method and an online early warning method based on an APO algorithm and a neural network. The prediction model training method comprises the following steps: extracting student features from a training sample; constructing a student learning condition prediction model, and determining an upper limit value and a lower limit value of a to-be-optimized parameter based on expert knowledge; initializing a student learning condition prediction model based on the values of the to-be-optimized parameters; training the student learning condition prediction model by using the training sample to obtain a trained student learning condition prediction model; and obtaining student characteristics of a to-be-evaluated student, and inputting the student characteristics into the trained student learning condition prediction model to obtain a learning condition of the student. The method improves the prediction performance and generalization ability of the model, and facilitates the improvement of the analysis accuracy of the learning condition of the student.

Description

Technical Field

[0001] The present invention relates to the technical field of data mining in the education industry, and in particular to a student learning situation prediction model training method and an online early warning method based on an APO algorithm and a neural network. Background Art

[0002] In today's higher education field, academic evaluation is the core link in ensuring the quality of education. Traditional academic evaluation technology has many limitations. It is not accurate in reflecting students' skills and knowledge mastery, and it is difficult to achieve personalized evaluation.

[0003] When using deep learning to predict student grades, most existing methods are limited to qualitative analysis at the specific course level. The depth of mining the grade data and the individual status of students is insufficient. In the face of nonlinear data, their prediction ability is weak and they cannot accurately describe the characteristics of student grade changes. At the same time, existing research has not deeply examined the learning motivation system at the subjective level of students, lacks analysis of the characteristics of student grade changes from the perspective of learning behavior and academic achievement, and is difficult to identify the key factors that affect students' academic performance, resulting in the inability to accurately warn students of their future learning status and to conduct effective intervention. Therefore, in-depth analysis of the main factors affecting academic performance and the construction of a scientific and effective academic warning model and mechanism have become key issues that need to be urgently addressed in the field of academic warning. Summary of the invention

[0004] The present invention proposes a student learning situation prediction model training method and an online early warning method based on the APO algorithm and neural network, which can solve the technical problem of early warning of the student's learning status, thereby early identification and intervention when the student has academic difficulties.

[0005] In each method embodiment of the present invention, an online early warning method for the learning situation of students based on the APO algorithm and neural network includes: Step S1: Construct a training sample based on the personal information of students, the historical data of students, and a number of image data randomly obtained during class by students within a preset time period; extract the student age, years of enrollment, academic performance index value, learning behavior index value, mental health index value, comprehensive quality and development potential index value, and human body posture characteristics from the training sample to form student characteristics; Step S2: Construct a prediction model for the learning situation of students. The prediction model for the learning situation of students includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence; determine the upper limit value and the lower limit value of the parameters to be optimized based on expert knowledge. The parameters to be optimized include the number of training times, learning rate, data volume size for each training, number of CNN filters, convolutional kernel size, regularization size in the student learning situation prediction model, and the number of LSTM hidden layer units in the BiGRU network module; use the APO algorithm to determine the values of the parameters to be optimized based on the upper limit value and the lower limit value of the parameters to be optimized; initialize the student learning situation prediction model based on the values of the parameters to be optimized; Step S3: Use the training sample to train the student learning situation prediction model to obtain a trained student learning situation prediction model.

[0006] Optionally, in the step S1, the academic performance index value is comprehensively quantified and generated based on the average course grade point, the number of failed courses, the number of retaken courses, the total semester grade, and the comprehensive credit of the make-up courses; the learning behavior index value is comprehensively quantified and generated based on the classroom participation degree, the homework completion degree, the independent learning duration, the learning method evaluation value, and the extracurricular learning activity participation degree; the mental health index value is comprehensively quantified and generated based on the emotion quantification value, the stress coping ability quantification value, and the mental health evaluation value; the comprehensive quality and development potential index value is comprehensively quantified and generated based on the professional skill mastery degree quantification value, the scientific research practice experience quantification value, the innovation activity participation quantification value, the campus culture and club activity participation quantification value, the volunteer service and social practice quantification value, and the leadership ability quantification value.

[0007] Optionally, an initial training sample set is composed of the training samples. For the original training samples in the initial training sample set with an equilibrium degree lower than the first preset threshold, the SMOTE algorithm is used to generate new training samples, and the original training samples are replaced by the new training samples to obtain a training sample set.

[0008] Optionally, in step S2, the student learning situation prediction model includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence, where: the convolutional neural network performs a convolution operation on the input student features, and the obtained convolution features are input into the BiGRU network module. The forward GRU module in the BiGRU network module obtains a first feature vector, and the backward GRU module obtains a second feature vector. The attention module adds an attention mechanism to the first feature vector and the second feature vector respectively to obtain weighted feature vectors, inputs the weighted feature vectors into the fully connected module, outputs the features to be classified, and inputs the features to be classified into the classification module to obtain a classification result; the classification result includes a normal state and an abnormal state.

[0009] Optionally, in step S2: using the APO algorithm to determine the value of the parameter to be optimized based on the upper limit value and the lower limit value of the parameter to be optimized, including: improving the APO algorithm using the reverse learning strategy of lens imaging. Among them, the reverse solution of the updated position is calculated using equation (15), and the protozoa are evaluated and updated using equation (16):

[0010] (15),

[0011] (16),

[0012] Among them, 、 respectively represent the original position, updated position, and reverse solution of the updated position of the th protozoa, represents the fitness function, lb i 、ub i represent the upper and lower limits of the search interval along the x-axis in the two-dimensional coordinates of the th protozoa, represents the scaling factor based on the lens imaging principle, k = 2sin(rand), and rand is a random number in the interval.

[0013] In each method embodiment of the present invention, an online early warning method for students' learning situation based on the APO algorithm and neural network is characterized in that the online early warning of students' learning situation is carried out using the student learning situation prediction model determined by the above method. The online early warning method for students' learning situation includes the following steps: Step S4: Obtain the student features of the student to be evaluated, and input the student features into the trained student learning situation prediction model to obtain the learning situation of the student. Among them, in step S4: obtaining the student features of the student to be evaluated includes: obtaining the data of the student to be evaluated, cleaning the student's data, removing duplicate data, deleting data with missing factors, repairing the variable type, and processing outliers.

[0014] In the above method embodiments of the present invention, a training device for a prediction model of students' learning situation based on the APO algorithm and neural network includes: a feature extraction module: configured to construct a training sample based on students' personal information, students' historical data, and a number of image data randomly obtained during class by students within a preset time period; extract students' age, years of enrollment, academic performance index values, learning behavior index values, mental health index values, comprehensive quality and development potential index values, and human body posture features from the training sample to form students' features; a model construction module: configured to construct a prediction model of students' learning situation, where the prediction model of students' learning situation includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence; determine the upper limit value and lower limit value of the parameters to be optimized based on expert knowledge, where the parameters to be optimized include the number of training times, learning rate, data volume size for each training, number of CNN filters, convolutional kernel size, regularization size in the prediction model of students' learning situation, and the number of LSTM hidden layer units in the BiGRU network module; use the APO algorithm to determine the values of the parameters to be optimized based on the upper limit value and lower limit value of the parameters to be optimized; initialize the prediction model of students' learning situation based on the values of the parameters to be optimized; a training module: configured to use the training sample to train the prediction model of students' learning situation to obtain a trained prediction model of students' learning situation.

[0015] In the above method embodiments of the present invention, an online early warning device for students' learning situation based on the APO algorithm and neural network uses the prediction model of students' learning situation determined by the above-mentioned training device for the prediction model of students' learning situation based on the APO algorithm and neural network to perform online early warning of students' learning situation. The online early warning device for students' learning situation includes: a calculation module: configured to obtain the students' features of the students to be evaluated, input the students' features into the trained prediction model of students' learning situation, and obtain the learning situation of the students.

[0016] In the above method embodiments of the present invention, a computer-readable storage medium stores multiple instructions, and the multiple instructions are used to be loaded and executed by a processor to perform the method as described above.

[0017] In the above method embodiments of the present invention, an electronic device includes: a processor for executing multiple instructions; a memory for storing multiple instructions; wherein, the multiple instructions are used to be stored by the memory and loaded and executed by the processor to perform the method as described above.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Description of the Drawings

[0019] The embodiments of the present invention will be described in more detail with reference to the accompanying drawings. The above and other objects, features, and advantages of the present invention will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 Schematic diagram of the training method for the prediction model of students' learning situation based on the APO algorithm and neural network of the present invention;

[0021] Figure 2 Schematic diagram of the process of the online warning method for students' learning situation based on the APO algorithm and neural network of the present invention;

[0022] Figure 3 Schematic diagram of the recurrent unit structure of the GRU network of the present invention;

[0023] Figure 4 Schematic diagram of the BiGRU network structure of the present invention;

[0024] Figure 5 Schematic diagram of the Attention mechanism of the present invention;

[0025] Figure 6 Schematic diagram of the APO algorithm of the present invention;

[0026] Figure 7 Schematic diagram of the interaction between the APO algorithm and the CNN - BiGRU - Attention model of the present invention;

[0027] Figure 8 Schematic diagram of the lens imaging reverse learning for improving the APO algorithm of the present invention;

[0028] Figure 9 Schematic diagram of the device for training the prediction model of students' learning situation based on the APO algorithm and neural network of the present invention;

[0029] Figure 10 Schematic diagram of the structure of the electronic device for online warning of students' learning situation based on the APO algorithm and neural network of the present invention. Detailed implementation manners

[0030] Next, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited by the exemplary embodiments described herein. It should be noted that: unless otherwise specifically stated, the relative arrangements, numerical expressions, and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present invention. Those skilled in the art can understand that terms such as "first", "second", S1, S2, etc. in the embodiments of the present invention are only used to distinguish different steps, devices, or modules, etc., and neither represent any specific technical meaning nor indicate an inevitable logical order between them. It should also be understood that in the embodiments of the present invention, "a plurality of" may refer to two or more, and "at least one" may refer to one, two, or more. It should also be understood that for any component, data, or structure mentioned in the embodiments of the present invention, without explicit limitation or contrary indication in the context, it can generally be understood as one or more. Additionally, the term "and / or" in the present invention is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in the present invention generally represents an "or" relationship between the associated objects before and after. It should also be understood that the present invention emphasizes the differences between the various embodiments, and their similarities or similarities can be referred to each other. For the sake of brevity, they will not be elaborated one by one. At the same time, it should be understood that for the sake of description, the dimensions of the various parts shown in the drawings are not drawn in actual proportional relationships. The following description of at least one exemplary embodiment is actually merely illustrative and in no way constitutes a limitation on the present invention and its application or use. Technologies, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and devices should be regarded as part of the specification. It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0031] Embodiments of the present invention can be applied to electronic devices such as terminal devices, computer systems, servers, etc., which can operate together with many other general or special computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments, and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, servers, etc. include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network personal computers, minicomputer systems, mainframe computer systems, and distributed cloud computing technology environments including any of the above systems, and so on. Electronic devices such as terminal devices, computer systems, servers, etc. can be described in the general context of computer system-executable instructions (such as program modules) executed by a computer system. Generally, program modules can include routines, programs, target programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. The computer system / server can be implemented in a distributed cloud computing environment where tasks are executed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can be located on local or remote computing system storage media including storage devices.

[0032] Exemplary method

[0033] As Figure 1 shown, a training method for a prediction model of students' learning situation based on the APO algorithm and neural network includes the following steps:

[0034] Step S1: Construct a training sample based on students' personal information, students' historical data, and several image data randomly obtained during class by students within a preset time period; extract students' age, years of enrollment, academic performance index values, learning behavior index values, mental health index values, comprehensive quality and development potential index values, and human body posture features from the training sample to form students' features;

[0035] Step S2: Construct a prediction model for students' learning situation, where the prediction model for students' learning situation includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence; determine the upper limit value and lower limit value of the parameters to be optimized based on expert knowledge, and the parameters to be optimized include the number of training times, learning rate, amount of data per training, number of CNN filters, convolutional kernel size, regularization size in the prediction model for students' learning situation, and the number of LSTM hidden layer units in the BiGRU network module; use the APO algorithm to determine the values of the parameters to be optimized based on the upper limit value and lower limit value of the parameters to be optimized; initialize the prediction model for students' learning situation based on the values of the parameters to be optimized;

[0036] Step S3: Use the training samples to train the prediction model for students' learning status, and obtain the trained prediction model for students' learning status.

[0037] Furthermore, an online early warning method for students' learning status based on the APO algorithm and neural network uses the prediction model for students' learning status determined by the above method to conduct online early warning of students' learning status. The online early warning method for students' learning status includes the following steps:

[0038] Step S4: Obtain the student characteristics of the student to be evaluated, and input the student characteristics into the trained prediction model for students' learning status to obtain the learning status of the student.

[0039] As an emerging technology for processing neural networks, deep learning provides a new solution for academic evaluation. Deep learning technology simulates the human neural network by constructing complex neural network models and natural language processing technologies. Through a multi-layer network structure, it captures and learns the complex patterns and relationships in the learning process of students; realizes the processing and analysis of complex data; and through image recognition, speech recognition and natural language processing of students' classes, homework, papers and test papers, realizes the accurate evaluation of students. In particular, the optimization algorithm in deep learning adjusts the model parameters, adaptively adjusts the learning rate, minimizes the loss function, improves the training efficiency and performance of the model, effectively conducts data analysis and pattern recognition, and significantly improves the performance of the automatic comment generation system through steps such as data preprocessing, model construction, and algorithm optimization, conducts an efficient, fair and comprehensive evaluation of the academic status of students, provides more accurate analysis and prediction; realizes more personalized and accurate teaching feedback, thereby promoting the personalized development of students and the improvement of teaching quality.

[0040] The prediction model for students' learning status of the present invention is a deep learning architecture that combines the characteristics of a convolutional neural network (CNN), a bidirectional gated recurrent unit (BiGRU) and an attention mechanism (Attention). This model uses CNN to capture the local features of the input data, uses BiGRU to obtain the context information of the sequence, and the attention mechanism is used to highlight important information and suppress unimportant information, so that the model can better understand and process the complex relationships in students' academic data. Further, the APO algorithm is also used to optimize the parameters of the prediction model for students' learning status, so as to further improve the prediction accuracy of the model.

[0041] Furthermore, in step S1, where:

[0042] The academic performance index value is comprehensively quantified and generated based on the average course credit grade point, the number of failed courses, the number of retaken courses, the total semester score, and the credit of make-up courses;

[0043] The learning behavior index value is comprehensively quantified and generated based on classroom participation, homework completion, self-study duration, learning method evaluation value, and extracurricular learning activity participation.

[0044] The mental health index value is comprehensively quantified and generated based on emotional quantification value, stress coping ability quantification value, and mental health evaluation value.

[0045] The comprehensive quality and development potential index value is comprehensively quantified and generated based on the quantification value of professional skill mastery, scientific research practice experience quantification value, innovation activity participation quantification value, campus culture and club activity participation quantification value, volunteer service and social practice quantification value, and leadership ability quantification value.

[0046] (1) Academic performance index value

[0047] Cumulative Grade Point Average (GPA): Reflects the overall academic performance of the trainee's mastery of professional knowledge.

[0048] Number of failed courses: Directly reflects the learning difficulties of the trainee in specific courses.

[0049] Number of repeated courses: Reflects the trainee's attention to failed courses and the effectiveness of remedial measures.

[0050] Total semester score: Evaluates the overall performance of the trainee in a single semester.

[0051] Credit of make-up courses: The number of credits of make-up courses can reflect the remedial situation of the trainee in academics.

[0052] (2) Learning behavior index value

[0053] Classroom participation: Includes interactions such as classroom questions and discussions, evaluates the trainee's classroom enthusiasm, and reflects the trainee's enthusiasm, interactivity, and learning attitude in class.

[0054] Homework completion: Reflects the trainee's attention to and completion of after-class learning tasks.

[0055] Self-study duration: Evaluates the time investment of the trainee in extracurricular self-study, as well as the initiative and self-discipline in learning.

[0056] Learning method evaluation value: Whether the trainee masters effective learning methods and strategies, such as time management, note-taking, etc.

[0057] Extracurricular learning activity participation: The number of times participating in extracurricular activities such as self-study groups, academic competitions, and research projects.

[0058] (3) Mental health index value

[0059] Emotional quantification value: The emotional stability of the trainee and the frequency of positive emotions.

[0060] Quantitative value of stress coping ability: Coping strategies for academic and life stress, such as seeking help, self-regulation, positive coping, negative coping and other stress coping methods.

[0061] Mental health assessment value: Evaluate the mental health status of students through professional mental health assessment tools, such as anxiety, depression, etc.

[0062] (4)Index values of comprehensive quality and development potential

[0063] Degree of professional skill mastery: Evaluate the students' mastery of professional knowledge through experiments, practical courses, etc.

[0064] Research practice experience: Experiences such as participating in research projects and laboratory work, reflecting the students' research ability.

[0065] Participation in innovation activities: Participate in innovation competitions, projects, etc., to evaluate the students' innovation ability and practical ability.

[0066] Participation in campus culture and club activities: The frequency and depth of participating in campus culture activities and club activities, reflecting the degree of integration of students into campus culture and social skills.

[0067] Volunteer service and social practice: Participate in volunteer service and social practice activities, reflecting the students' interests and hobbies, and evaluating the students' sense of social responsibility.

[0068] Leadership ability: Evaluate the leadership ability and organizational coordination ability of students in team projects and collective activities.

[0069] Through these multi-dimensional index values, the present invention can comprehensively and accurately evaluate the academic status of students, timely discover potential academic risks, so as to achieve early intervention and precise assistance, and improve the personal development of students.

[0070] Human body posture characteristics can reflect the listening state of students during class, which helps to improve the accuracy of analysis.

[0071] Furthermore, an initial training sample set is composed of training samples. For the original training samples in the initial training sample set whose balance degree is lower than the first preset threshold, the SMOTE algorithm is used to generate new training samples, and the original training samples are replaced by the new training samples to obtain a training sample set.

[0072] In the present invention, the basic idea of the SMOTE algorithm is to conduct a detailed analysis of the minority class samples, and synthetically generate new samples according to the minority class samples and add them to the data set. In this way, the balance of the data set is improved, and the generalization ability of the model is enhanced.

[0073] (1) For each sample x in the minority class, calculate the distance between the point and other sample points in the minority class and obtain the nearest k neighbors (i.e., perform the KNN algorithm on the minority class points).

[0074] (2) A sampling ratio is set according to the sample imbalance ratio to determine the sampling rate. For each minority class sample x, several samples are randomly selected from its k nearest neighbors. Assume that the selected nearest neighbor is x'.

[0075] (3) For each randomly selected neighbor x', construct a new sample x with the original sample according to the following formula: new :

[0076]

[0077] Furthermore, in step S2, the student learning situation prediction model includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module and a classification module connected in sequence, wherein:

[0078] The convolutional neural network performs a convolution operation on the input student features, and the obtained convolution features are input into the BiGRU network module. The forward GRU module in the BiGRU network module obtains the first feature vector, and the reverse GRU module obtains the second feature vector. The attention module adds an attention mechanism to the first feature vector and the second feature vector respectively to obtain a weighted feature vector, which is input into the fully connected module and the features to be classified are output. The features to be classified are input into the classification module to obtain the classification results; the classification results include normal state and abnormal state.

[0079] In the present invention, the convolutional neural network module, BiGRU network module, attention module, fully connected module and classification module all use conventional structures in the art.

[0080] like Figures 2 - 7 As shown in the figure, CNN (convolutional neural network) is a feedforward neural network that can mine the deep features contained in discriminative data through alternating convolutional layers and pooling layers. Each neuron in the fully connected layer is fully connected to the neurons in the previous layer, and can integrate the features extracted by the convolutional layer and the pooling layer to obtain more discriminative features. Through a series of convolutional and pooling operations, the relevant features in the input data are gradually extracted, and the prediction calculation is completed through the fully connected layer, thus having powerful feature extraction and recognition capabilities.

[0081] For subsequent pattern recognition and classification calculations, the results need to be converted into probabilistic form. The input data belongs to The probability value S of the quality-like mode j It can be calculated by the softmax function:

[0082]

[0083] ,

[0084]

[0085]

[0086] Wherein: is the output value of the th neuron on the th feature map in the first convolutional layer; is the ReLU activation function; is the th weight value of the th row and th column in the th convolutional kernel; is the input data; is the bias value corresponding to the th convolutional kernel; is the concatenated feature vector; is the connection weight between the th neuron in the fully connected layer and the th element of the input vector; is the bias value of the th neuron in the fully connected layer; is the number of neurons in the fully connected layer; is the connection weight between the th neuron in the output layer and the th neuron in the fully connected layer; is the bias value of the th output layer neuron;

[0087] The GRU (Gated Recurrent Unit) network is a variant of the long short-term memory network, with advantages such as simple structure, few network parameters, high computational efficiency, and small storage space. It can effectively alleviate the problems of gradient disappearance or gradient explosion, thereby capturing the dependencies between data. Based on the overall prediction of the sequence output, it has the ability to learn using the information contained in the context of the sequence data. The BiGRU network contains two GRU layers, the forward and backward ones. The forward-propagating GRU is used to calculate the sequence information at the current moment, and the backward-propagating GRU reads the same sequence in reverse, introducing reverse-order information. The two GRU network layers are jointly connected to an output layer, providing both forward and reverse-order information for all neurons in the output layer during the network training process.

[0088] The BiGRU network performs calculations independently through equations (01)-(02), with no interaction between networks. Each network updates its state and generates an output, and finally, the outputs in two directions are concatenated according to equation (03).

[0089] (01)

[0090] (02)

[0091] (03)

[0092] In the equations: 、 、 、 are the forward hidden layer state, backward hidden layer state, input value of the input neuron, and output value of the hidden layer state at time 、 、 、 weight matrices of different components; is the activation function of the hidden layer; 、 are the bias vectors of the forward hidden layer and backward hidden layer; is the hidden layer state vector concatenation operation.

[0093] Attention (attention mechanism) is a model in deep learning based on the human visual nervous system. This model assigns corresponding weights to the hidden layer vectors of the input sequence at different times, achieving the effect of reducing interference from irrelevant information and improving the prediction accuracy of the model.

[0094]

[0095]

[0096]

[0097]

[0098] In the equations: is a vector sequence containing elements x; is a query vector; is the weight of the th element; is the attention scoring function, used to measure the similarity between and the query vector ; is the attention vector.

[0099] The APO (Artificial Protozoa Optimizer) simulates the survival mechanism of protozoa by mimicking the foraging, dormancy, and reproduction behaviors of protozoa. The functions of bacteria, algae, and protozoa in microorganisms are similar to the organs of higher animals and plants, and they are accomplished through special structures called "organelles". These microorganisms exhibit basic life characteristics, including metabolism, reproduction, genetic continuity, variability, and adaptation to environmental stimuli. Microorganisms are usually utilized more effectively than higher organisms because they have simpler tissues and lower complexity. Protozoa refer to representative true algae among flagellates. This algorithm features fewer parameters, a fast convergence rate, strong search capabilities, and a wide range of applications.

[0100] According to the present invention, as described below, the lens imaging reverse learning strategy (IV) is further utilized to improve the principle algorithms (I) - (III) of the following APO.

[0101] (I) Foraging behavior

[0102] 1) Autotrophic mode

[0103] Protozoa can produce carbohydrates through chloroplasts to provide nutrition. If a protozoan is under strong light irradiation, it will move away from its current position towards a position with lower light intensity. Conversely, if it is in a position with low light intensity, it will move towards a position with higher light intensity. Assume that the light intensity around the th protozoan is suitable for photosynthesis, then this protozoan will move to the position where the th protozoan is located. For the autotrophic mode, the mathematical model includes the following equations (1) - (6):

[0104] (1)

[0105] (2)

[0106] (3)

[0107] (4)

[0108] (5)

[0109] (6)

[0110] Where: and respectively represent the updated position and the original position of the th protozoan; is the th protozoan randomly selected; Indicates a protozoan whose ranking index randomly selected from among the neighbors is less than ; specifically, if is , then is also set to ; Indicates a protozoan whose ranking index randomly selected from among the neighbors is greater than ; in particular, if is (where is the population size), then is also set to ; Indicates the foraging factor, : fitness function; rand represents a number randomly selected from a uniform distribution in the interval [0, 1]; iter and iter max represent the current iteration number and the maximum iteration number, respectively; Indicates the number of neighbor pairs in the external factors, and is 's maximum value; is a weight factor in the autotrophic mode; eps (2.2204e - 16) is an extremely small number; ⊙ represents the Hadamard product; is a foraging mapping vector of size (1 × ), where each element is 0 or 1. di represents the dimension index, and , dim: the number of decision variables; sort(.): sorts in ascending order according to the fitness value; randperm( : returns a row vector containing randomly selected unique integers between 1 and ; : ceiling function, : floor function.

[0111] 2) Heterotrophic mode

[0112] In the dark, protozoa can obtain nutrients by absorbing organic matter in the surrounding environment. Assume there is a food - rich location nearby, and the protozoa will move towards it. For the heterotrophic mode, the following mathematical model is proposed, including the following equations (7) - (10):

[0113] (7)

[0114] (8)

[0115] (9)

[0116] (10)

[0117] Wherein: " " represents a nearby position; " " means that the position can extend in different directions starting from the th protozoan. represents the th protozoan selected from the nd pair of neighbors, and its ranking index is . Specifically, if is , then is also set to . represents the th protozoan selected from the nd pair of neighbors, and its ranking index is . In particular, if is , then is also set to . is a weight factor in the heterotrophic mode. Rand is a random vector, and its element rand is a random number within the interval.

[0118] (II) Dormancy

[0119] Under environmental stress conditions, protozoa may adopt dormancy behavior as a survival strategy to endure adverse conditions. When a protozoan is in a dormant state, it will be replaced by newly generated protozoa to maintain the stability of the population size. The mathematical model of dormancy includes the following equations (11) - (12):

[0120] (11)

[0121] (12)

[0122] Wherein: and represent the lower bound vector and the upper bound vector respectively. and represent the lower bound and the upper bound of the th variable respectively.

[0123] (III) Reproduction

[0124] Under appropriate age and health conditions, protozoa reproduce asexually, a process known as binary fission. In theory, this mode of reproduction causes the protozoa to split into two identical offspring. This behavior is simulated by generating a copy identical to the protozoa and considering a perturbation. The mathematical model of reproduction includes the following equations (13) - (4)::

[0125] (13)

[0126] (14)

[0127] where: " " indicates that the perturbation can be positive or negative. is a mapping vector in the reproduction process, with a magnitude of and each element being 0 or 1. The proposed Adaptive Prototype Optimization (APO) algorithm has two special parameters: (number of neighbor pairs) and (maximum proportion fraction).

[0128] The parameters involved in the APO algorithm are as follows:

[0129] rand

[0130]

[0131]

[0132] where: is the proportion fraction of dormancy and reproduction in the protozoa population. is 's maximum value. represents the probabilities of autotrophic and heterotrophic behaviors. represents the probabilities of dormancy and reproduction.

[0133] (IV) Lens Imaging Reverse Learning Strategy

[0134] The main idea of reverse learning is to generate a reverse position based on the current coordinates to expand the search range. This can not only jump out of the current position but also expand the search range, improving the diversity of the population and thus enhancing the accuracy of the training model and the prediction accuracy.

[0135] As known from Figure 8 , in two-dimensional coordinates, the search range of the x-axis is The y-axis represents a convex lens. Assume 's projection on the x-axis is , and the height is , and the imaging on the other side can be obtained through lens imaging as , The projection on the x-axis is , and the height is . Through the above principle, the inverse projection of can be obtained .

[0136] Individual gets its corresponding inverse point with o as the base point . According to the lens imaging principle, it can be obtained that:

[0137]

[0138] Denote as the scaling factor, and the inverse learning formula based on the lens imaging principle can be obtained:

[0139] (15)

[0140] Since the fixed scaling factor cannot fully utilize the advantages of the lens imaging learning strategy, a dynamic lens imaging learning strategy is proposed, and let .

[0141] , respectively represent the original position, updated position, and inverse solution of the updated position of the th protozoan. When the fitness value of the updated position or the inverse solution of the updated position is better than that of the original individual, it is replaced; otherwise, it is not replaced. The formula is as follows:

[0142] (16)

[0143] In summary, the APO algorithm process improved by the lens imaging inverse learning strategy includes the following steps:

[0144] S11: Initialize parameters

[0145] S12: Evaluate the maximum fitness;

[0146] S13: Sort according to the fitness value and distinguish whether it is foraging behavior, dormancy or reproduction, and judge:

[0147] 1) If it is foraging behavior, judge according to whether it is autotrophic mode:

[0148] a) If it is autotrophic mode, generate new protozoa using equation (1);

[0149] b) If it is not autotrophic mode, generate new protozoa using equation (7);

[0150] 2) If it is dormancy or reproduction behavior, judge:

[0151] c) If it is a dormant behavior, use Equation (11) to generate new protozoa;

[0152] d) If it is a reproductive behavior, use Equation (14) to generate new protozoa;

[0153] S14: Use the lens imaging reverse learning strategy to update the protozoa using Equations (15) and (16);

[0154] S15: Evaluate and update the protozoa;

[0155] S16: Update the global best result;

[0156] S17: Return to step S12. If the requirements are not met, repeat steps S13 - S16. If the requirements are met, end the process.

[0157] In this way, the lens imaging reverse learning strategy can be used to improve the APO algorithm. Among them, the lens imaging reverse learning strategy can be used to update the protozoa using Equations (15) and (16), thereby further optimizing the parameters of the student learning situation prediction model and further improving the model prediction accuracy:

[0158] (15),

[0159] (16),

[0160] Among them, 、 respectively represent the original position, updated position, and reverse solution of the updated position of the th protozoa, represents the fitness function, lb i 、ub i represent the upper and lower limits of the search interval along the x-axis in the two-dimensional coordinates of the th protozoa, represents the scaling factor based on the lens imaging principle, k = 2sin(rand), where rand is a random number in the interval. Among them, the reverse solution of the updated position is calculated using Equation (15), and the protozoa are evaluated and updated using Equation (16).

[0161] The present invention combines the advantages of CNN, BiGRU, and Attention to construct a student learning situation prediction model, and uses the improved APO algorithm to optimize the parameters of the student learning situation prediction model: the number of training times, batch_size (the size of the data volume for each training), learning rate, the number of CNN filters, filter kernel size, regularization parameter, and the number of LSTM hidden layer units.

[0162] Step S4: Obtain the characteristics of the trainee to be evaluated, including:

[0163] Obtain the data of the trainee to be evaluated, clean the trainee's data, remove duplicate data, delete data with missing factors, repair the variable type, and handle outliers.

[0164] By comprehensively checking and handling various problems in the academic data of trainees, de-duplicate the duplicate data, delete the data with missing factors, repair the incorrect variable type, and use reasonable methods to handle outliers. For example, for data with abnormal fluctuations in grades, verify and correct it in combination with the actual situation of the trainee to ensure data quality.

[0165] In this application, when extracting features from the training samples:

[0166] (1) For numerical indicators, since the dimensions and value ranges of each numerical indicator vary greatly, in order to eliminate the influence of the characteristics of each indicator on the model due to different dimensions and value range differences, perform a normalization transformation on the numerical indicators. Use the min-max normalization method, formula:

[0167]

[0168] where and are the maximum and minimum values of the numerical indicator respectively. Through the normalization transformation, map the data of the numerical indicator characteristics to the interval to make different indicators comparable.

[0169] (2) For categorical indicators, since each categorical indicator does not have sequence and cannot be compared in size, it cannot be replaced by a numerical value because the numerical size will affect the calculation of the weight matrix, and there is no size relationship attribute, and its weight should not change accordingly. To ensure the accuracy of the weight matrix calculation, one-hot encoding is required. Suppose the indicator has a total of different states. When the indicator of the sample is the th state, then the th position is 1, and the remaining positions are all 0. The one-hot encoding of this sample is represented as .

[0170] Divide the features corresponding to the training samples into a training set and a validation set according to 8:2.

[0171] By reasonably dividing the data processed by feature engineering into a training set and a validation set according to a certain proportion, it provides data support for the subsequent training and evaluation of the model. Use the validation set to evaluate the trained prediction model of the trainee's learning situation.

[0172] The evaluation metrics are accuracy, precision, recall, and F1 score. Denote: Indicates predicting the positive class of the label as the positive class, Indicates predicting the negative class of the label as the positive class, Indicates predicting the positive class of the label as the negative class, Indicates predicting the negative class of the label as the negative class. Then the evaluation metrics are:

[0173] (1) Accuracy

[0174]

[0175] (2) Precision

[0176]

[0177] (3) Recall

[0178]

[0179] (4) F1 score

[0180]

[0181] Furthermore, obtain the student characteristics of the student to be evaluated, input the student characteristics into the trained student learning situation prediction model, obtain the learning situation of the student, and give an early warning for the student's academic performance.

[0182] The present invention adopts the following technical means:

[0183] 1. Handling imbalanced datasets: In academic early warning, the amount of data in some academic performance categories (such as excellent grades) may be much larger than that in other categories (such as failing grades), resulting in an imbalanced dataset. By using oversampling, undersampling, or algorithm-level optimization to handle the imbalanced dataset problem, the generalization ability of the model is improved.

[0184] 2. Feature engineering: After preprocessing the learning data of students, such as cleaning, deduplicating, transforming, and encoding, use correlation analysis and data mining techniques to screen out the features most relevant to academic performance, explore the laws and characteristics of students' academic problems, and select appropriate features to improve the accuracy of the model.

[0185] 3. Model construction and optimization: On the basis of fully analyzing the advantages of machine learning algorithms APO, CNN, BiGRU, and Attention, construct a student learning situation prediction model, and perform training and optimization. This mainly includes hyperparameter tuning, model selection, and cross-validation, etc., to improve the prediction performance and generalization ability of the model.

[0186] 4. Model evaluation and application: The model is evaluated using the test data set. Based on the results predicted by the model, the academic performance of the students is intervened and adjusted, and early warnings and intervention strategies are provided for students who may need additional help to improve the academic achievements of the students.

[0187] 5. Real-time monitoring and dynamic early warning: It can update data in real time and respond quickly, dynamically monitor and give early warnings about the learning status of students; it has efficient data processing capabilities and real-time analysis capabilities.

[0188] Exemplary apparatus

[0189] Figure 9 It is a schematic structural diagram of a training device for a student learning situation prediction model based on the APO algorithm and neural network provided by an exemplary embodiment of the present invention. As Figure 9 shown, this device includes:

[0190] Feature extraction module: Configured to construct training samples based on the personal information of students, the historical data of students, and several image data randomly obtained during class by students within a preset time period; extract the student age, years of enrollment, academic performance index values, learning behavior index values, mental health index values, comprehensive quality and development potential index values, and human body posture features from the training samples to form student features;

[0191] Model construction module: Configured to construct a student learning situation prediction model. The student learning situation prediction model includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence; determine the upper limit value and lower limit value of the parameters to be optimized based on expert knowledge. The parameters to be optimized include the number of training times, learning rate, data volume size per training, number of CNN filters, convolutional kernel size, regularization size in the student learning situation prediction model, and the number of LSTM hidden layer units in the BiGRU network module; use the APO algorithm to determine the values of the parameters to be optimized based on the upper limit value and lower limit value of the parameters to be optimized; initialize the student learning situation prediction model based on the values of the parameters to be optimized;

[0192] Training module: Configured to train the student learning situation prediction model using the training samples to obtain a trained student learning situation prediction model.

[0193] Furthermore, an online early warning device for the learning situation of students based on the APO algorithm and neural network uses the student learning situation prediction model determined by the above-mentioned student learning situation prediction model training device based on the APO algorithm and neural network to perform online early warning of the student learning situation. The online early warning device for the student learning situation includes: a calculation module: configured to obtain the student characteristics of the student to be evaluated, input the student characteristics into the trained student learning situation prediction model, and obtain the learning situation of the student. In the example as Figure 7 shown, the prediction result can be obtained by acquiring the online academic data of the student, processing it through feature engineering, and then inputting it into the model for prediction.

[0194] Exemplary electronic device

[0195] Figure 10 is the structure of the electronic device 90 provided by an exemplary embodiment of the present invention. The electronic device can be any one or both of the first device and the second device, or a stand-alone device independent of them. The stand-alone device can communicate with the first device and the second device to receive the input signals collected from them. Figure 10 The block diagram of the electronic device according to an embodiment of the present disclosure is illustrated. As Figure 10 shown, the electronic device includes one or more processors 91 and a memory 92.

[0196] The processor 91 can be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and can control other components in the electronic device to perform desired functions.

[0197] The memory 92 can include one or more computer program products, and the computer program products can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory can include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory can include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions can be stored on the computer-readable storage media, and the processor 91 can run the program instructions to implement the methods of the software programs of the various embodiments of the present disclosure described above and / or other desired functions. In one example, the electronic device may further include: an input device 93 and an output device 94, and these components are interconnected through a bus system and / or other forms of connection mechanisms (not shown). In addition, the input device 93 may further include, for example, a keyboard, a mouse, etc. The output device 94 can output various information to the outside. The output device 94 can include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0198] Of course, for simplicity, Figure 10 only some of the components related to the present disclosure in the electronic device are shown, and components such as buses, input / output interfaces, etc. are omitted. In addition, according to specific application scenarios, the electronic device may further include any other appropriate components.

[0199] Exemplary computer program product and computer-readable storage medium

[0200] In addition to the above methods and devices, embodiments of the present disclosure may also be computer program products, which include computer program instructions that, when run by a processor, cause the processor to execute the steps in the methods according to various embodiments of the present disclosure described in the "Exemplary Methods" section above of this specification.

[0201] The computer program products may be written in any combination of one or more programming languages for programming code to perform the operations of the embodiments of the present disclosure. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's device, executed as a stand-alone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0202] Furthermore, embodiments of the present disclosure may also be computer-readable storage media, on which computer program instructions are stored, and the computer program instructions, when run by a processor, cause the processor to execute the steps in the methods according to various embodiments of the present disclosure described in the "Exemplary Methods" section above of this specification.

[0203] The computer-readable storage media may employ any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may, for example, include but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0204] The basic principles of the present disclosure have been described in conjunction with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present disclosure are merely examples and not limitations. It cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present disclosure. Additionally, the specific details disclosed above are only for illustrative and facilitating understanding purposes, and not for limitation. The above details do not limit the present disclosure to necessarily implementing with the above specific details.

[0205] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other. For system embodiments, since they basically correspond to method embodiments, the description is relatively simple. For related parts, reference can be made to the corresponding descriptions in the method embodiments.

[0206] The block diagrams of the devices, apparatuses, equipment, and systems involved in the present disclosure are only illustrative examples and do not intend to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any way. Words such as "including", "comprising", "having", etc. are open-ended terms, meaning "including but not limited to", and can be used interchangeably with each other. The word "or" and "and" used herein refer to the phrase "and / or", and can be used interchangeably with each other, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to", and can be used interchangeably with each other.

[0207] The methods and apparatuses of the present disclosure can be implemented in many ways. For example, the methods and apparatuses of the present disclosure can be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of the steps for the methods is only for illustration purposes. The steps of the methods of the present disclosure are not limited to the specific order described above, unless otherwise specifically stated in other ways. Additionally, in some embodiments, the present disclosure can also be implemented as a program recorded in a recording medium, and these programs include machine-readable instructions for implementing the methods according to the present disclosure. Therefore, the present disclosure also covers the recording medium storing the programs for executing the methods according to the present disclosure.

[0208] It should also be noted that in the devices, equipment and methods of the present disclosure, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations shall be regarded as equivalent solutions of the present disclosure. The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

[0209] The above description has been presented for purposes of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize some of their variations, modifications, alterations, additions, and subcombinations.

Claims

1. A training method for a prediction model of students' learning situation based on the APO algorithm and neural network, characterized in that The method includes the following steps: Step S1: Construct a training sample based on the personal information of the trainee, the historical data of the trainee, and several image data randomly obtained during the trainee's class in a preset time period; extract the trainee's age, years of enrollment, academic performance index value, learning behavior index value, mental health index value, comprehensive quality and development potential index value, and human body posture characteristics from the training sample to form the trainee characteristics; Step S2: Construct a trainee learning situation prediction model. The trainee learning situation prediction model includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence; determine the upper limit value and the lower limit value of the parameters to be optimized based on expert knowledge. The parameters to be optimized include the number of training times, learning rate, data volume size for each training, number of CNN filters, convolutional kernel size, regularization size in the trainee learning situation prediction model, and the number of LSTM hidden layer units in the BiGRU network module; use the APO algorithm to determine the values of the parameters to be optimized based on the upper limit value and the lower limit value of the parameters to be optimized; initialize the trainee learning situation prediction model based on the values of the parameters to be optimized; Step S3: Use the training sample to train the trainee learning situation prediction model to obtain the trained trainee learning situation prediction model.

2. The method according to claim 1, characterized in that, For the said Step S1, where: The academic performance index value is comprehensively quantified based on the average course GPA, the number of failed courses, the number of retaken courses, the total semester grade, and the credit of make-up courses; The learning behavior index value is comprehensively quantified based on classroom participation, assignment completion, autonomous learning duration, learning method evaluation value, and extracurricular learning activity participation; The mental health index value is comprehensively quantified based on the emotion quantification value, stress coping ability quantification value, and mental health evaluation value; The comprehensive quality and development potential index value is comprehensively quantified based on the quantification value of professional skill mastery, the quantification value of scientific research practice experience, the quantification value of innovation activity participation, the quantification value of campus culture and club activity participation, the quantification value of volunteer service and social practice, and the quantification value of leadership ability.

3. The method according to claim 1, wherein The initial training sample set is composed of the training samples. For the original training samples in the initial training sample set whose balance degree is lower than the first preset threshold, use the SMOTE algorithm to generate new training samples, and replace the original training samples with the new training samples to obtain the training sample set.

4. The method according to claim 1, characterized in that, For the said Step S2, the trainee learning situation prediction model includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence, where: The convolutional neural network performs a convolution operation on the input trainee characteristics, and the obtained convolution features are input into the BiGRU network module. The forward GRU module in the BiGRU network module obtains the first feature vector, and the backward GRU module obtains the second feature vector. The attention module adds an attention mechanism to the first feature vector and the second feature vector respectively to obtain a weighted feature vector. The weighted feature vector is input into the fully connected module, and the feature to be classified is output. The feature to be classified is input into the classification module to obtain the classification result; the classification result includes the normal state and the abnormal state.

5. The method according to claim 1, wherein The step S2: Using the APO algorithm to determine the value of the parameter to be optimized based on the upper limit value and the lower limit value of the parameter to be optimized, including: improving the APO algorithm by using the reverse learning strategy of lens imaging, where the reverse solution of the updated position is calculated using Equation (15), and the protozoa are evaluated and updated using Equation (16): （15）， （16）， Among them, and respectively represent the original position, updated position, and inverse solution of the updated position of the -th protozoan. represents the fitness function, lb i and ub i represent the upper and lower limits of the search interval along the x-axis in the two-dimensional coordinates of the -th protozoan. represents the scaling factor based on the lens imaging principle, k = 2sin(rand), where rand is a random number in the interval.

6. An online early warning method for the learning situation of students based on the APO algorithm and neural network, characterized in that, Using the prediction model of the learning situation of students determined by the method according to any one of Claims 1 to 5, to conduct online early warning of the learning situation of students, the online early warning method of the learning situation of students includes the following steps: Step S4: Obtain the student characteristics of the student to be evaluated, input the student characteristics into the trained prediction model of the learning situation of students, and obtain the learning situation of the student. Among them, the step S4: Obtain the student characteristics of the student to be evaluated, including: obtaining the data of the student to be evaluated, cleaning the data of the student, removing duplicate data, deleting the data with missing factors, repairing the variable type, and processing the outliers.

7. A training device for a prediction model of the learning situation of trainees based on the APO algorithm and neural network, characterized in that, The device includes: Feature extraction module: configured to construct a training sample based on the personal information of the student, the historical data of the student, and several image data randomly obtained during the class of the student within a preset time period; extract the student age, years of enrollment, academic performance index value, learning behavior index value, mental health index value, comprehensive quality and development potential index value, and human body posture characteristics from the training sample to form student characteristics; Model construction module: configured to construct a prediction model of the learning situation of students, the prediction model of the learning situation of students includes a convolutional neural network module, a BiGRU network module, an attention module, a fully connected module, and a classification module connected in sequence; determine the upper limit value and the lower limit value of the parameter to be optimized based on expert knowledge, and the parameters to be optimized include the number of training times, learning rate, amount of data per training, number of CNN filters, convolutional kernel size, regularization size in the prediction model of the learning situation of students, and the number of LSTM hidden layer units in the BiGRU network module; use the APO algorithm to determine the value of the parameter to be optimized based on the upper limit value and the lower limit value of the parameter to be optimized; initialize the prediction model of the learning situation of students based on the value of the parameter to be optimized; Training module: configured to use the training sample to train the prediction model of the learning situation of students to obtain the trained prediction model of the learning situation of students.

8. An online early warning device for the learning situation of students based on the APO algorithm and neural network, characterized in that, Using the prediction model of the learning situation of students determined by the training device of the prediction model of the learning situation of students based on the APO algorithm and neural network according to Claim 7, to conduct online early warning of the learning situation of students, the online early warning device of the learning situation of students includes: Calculation module: configured to obtain the student characteristics of the student to be evaluated, input the student characteristics into the trained prediction model of the learning situation of students, and obtain the learning situation of the student.

9. A computer-readable storage medium, in which multiple instructions are stored; the multiple instructions are used to be loaded and executed by a processor to execute the method according to any one of Claims 1 to 6.

10. An electronic device, characterized in that, The electronic device includes: A processor for executing multiple instructions; A memory for storing multiple instructions; Among them, the multiple instructions are used to be stored by the memory and loaded and executed by the processor to execute the method according to any one of Claims 1 to 6.

Images