Research On Academic Early Warning Model Based On Improved Cat Swarm Optimization Multi-channel Convolutional Neural Network

With the rapid development of network technology,database technology and other computer technology,the construction of digital campus has been widely used.Colleges and universities have applied information technology to educational management.Various application systems in colleges and universities have produced massive data after long-term use and operation.People begin to realize that educational data can not only be counted,stored,queried,added or modified in the simple stage of student information.How to rationally use artificial intelligence and big data technology to solve the problems existing in improving teaching quality and college students' education and teaching process has become a research hotspot in the field of education.Therefore,the use of data mining and artificial intelligence technology to build a reasonable student learning early warning model system,for improving the quality of teaching,training high-quality talents,in the teaching reform of colleges and universities is of great significance.The study of data mining reflects frequent patterns,associations,associations,or causality between item sets,and discovers special and valuable knowledge hidden in the database.Current researches mainly focus on improving algorithm efficiency,but ignore the understanding and participation of users.The human-computer interaction interface mode provided by some data mining applications is too monotonous,especially the mining process jumps too fast,which makes users unable to understand the mining results well.In order to improve the efficiency of students' academic warning,this article warning process is as follows:First,in this paper,by analyzing the education process of data mining,and the characteristics of the data of a normal university,selected the student's social characteristics,personal characteristics and student behavior from three aspects,a total of 13 of the evaluation index,and combined with the university regulations,divides the three warning level.The data of13 evaluation index came from the data of students' historical achievements in the educational administration system of a normal university,the data of the times of entering the library and the number of books borrowed from the campus card system,the data of students' basic information in the student management system and the data of students' participation in extracurricular activities and academic lectures in the electronic version.Secondly,data preprocessing operations such as data screening and data cleaning are carried out on the collected data,and a typical multi-channel convolutional neural network(MCNN)in deep learning technology is applied to warn students about their academic performance.In addition,optimize the network topology of CNN to improve the performance of the model.CNN has many hyperparameters that need to be adjusted to build an optimal model that can effectively learn data patterns.In this paper,an improved cat swarm optimization(CSO)is proposed to optimize the parameters of CNN model and further optimize the feature extraction part of CNN.In order to verify the effectiveness of the proposed model,the prediction results are compared with CSO-CNN models.The experimental results show that the CNN based on the improved cat swarm optimization is superior to the CNN based on the cat swarm optimization,which proves the effectiveness of the hybrid algorithm of improved CSO and CNN,and can effectively give early warning to the academic situation of college students.Finally,the problems existing in the application of the academic warning model are analyzed and corresponding suggestions are given.