Application Of Back Propagation Neural Network Based Teaching Cognitive Diagnosis

Under the modern education teaching concept of people-oriented and innovative talent training,educators are committed to making every learner get personalized learning experience and get full development.In our country's school education,students and teachers are relatively high,so it is difficult for teachers to pay attention to each student and provide personalized instruction.Computer-based cognitive diagnostic tests can provide individualized remedial learning strategies for students by diagnosing their cognitive structures.Cognitive diagnosis is the basis of individualized learning.Regular space model is one of the effective methods of cognitive diagnosis.But the regular space model usually requires a large number of samples for parameter estimation to obtain accurate diagnostic results,which is applicable to large exams with many students,complex knowledge structure,large amount of questions and long period,.However,In practical teaching,it is often necessary to carry out small sample diagnostic tests within a short period to timely remedy knowledge gaps of students,so as to avoid problems accumulation and promote personalized learning of students.In the study,on the basis of a large number of domestic and overseas related literature,the author found that the BP neural network pattern recognition method is not restricted by sample size.Therefore,this study combined the Q matrix theory of regular space model with the pattern recognition method of BP neural network and proposed back propagation neural network based cognitive diagnosis,so as to reduce the requirement of regular space model on the number of samples and realize the cognitive diagnosis of small samples.The basic idea of this method is to retain the Q matrix theory of regular space model and replace its large sample pattern recognition stage with a small sample BP neural network self-organizing,self-learning and self-adaptive pattern recognition method.Firstly,cognitive diagnostic attributes and test items are determined to generate data pairs of ideal attribute patterns and expected response patterns.Next,BP neural network is trained with this data as the sample to carry out cognitive diagnostic tests.Then,the trained neural network should be applied to pattern recognition.Finally,the cognitive diagnosis report is generated,the cognitive structure of students is analyzed,and targeted learning resources and learning paths are recommended to students based on the diagnosis results.This study is based on the course of the Java language programming.Back propagation neural network based cognitive diagnosis is applied to practical teaching to test its feasibility and effectiveness.The study conducted data collection and statistical analysis from four aspects: diagnostic result accuracy rate,diagnostic efficiency,student learning effect and student satisfaction.Finally,a conclusion is drawn: back propagation neural network based cognitive diagnosis has a high accuracy in small sample cognitive diagnosis and can accurately diagnose the knowledge structure of students.The diagnosis efficiency of this method is high and it has real time in teaching application.The diagnosis results of this method provide the basis for students' personalized remedial learning and can effectively improve their learning effect.Students are satisfied with the teaching application of this method.In conclusion,back propagation neural network based cognitive diagnosis is feasible and effective.