基于MLP-Bagging集成分类模型的在线学习行为分析  

Analysis of online learning behavior based on MLP-Bagging ensemble classification model

作　　者：普运伟 姜萤 田春瑾 余永鹏 PU Yunwei;JIANG Ying;TIAN Chunjin;YU Yongpeng(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,Yunnan,China;Computer Center,Kunming University of Science and Technology,Kunming 650500,Yunnan,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650500 [2]昆明理工大学计算中心,云南昆明650500

出　　处：《云南大学学报（自然科学版）》2024年第5期852-861,共10页Journal of Yunnan University(Natural Sciences Edition)

基　　金：教育部−思科产学研项目(201701010017);昆明理工大学课程思政教改专项(2021003);昆明理工大学课程思政示范课程项目(2021152008)。

摘　　要：针对教育者难以对学习者多样化的在线学习行为进行监测和研判等问题,提出一种带嵌入层的MLP-Bagging集成分类模型对学习者的在线学习行为进行分析与判别.考虑到学习者的在线学习行为以及个体特性,从学习准备行为、知识获取行为、交互学习行为、学习巩固行为和辅助特征5个方面构建在线学习行为模型,并采用MLP-Bagging集成分类模型对学习者进行分类判别.实验结果表明,所构建的学习模型可对在线学习者的学习行为进行符合实际的建模,加入辅助特征有利于对各类学习者的在线学习行为进行深入的分析与指导,并且在分类模型中加入嵌入层可以有效克服标签编码带来的数据冗余和误差缺陷,从而获得更好的分类效果.与其他分类模型相比,融合多个MLP分类器的Bagging集成模型可以减少单个MLP分类器的方差,其分类准确率达到98.72%,具有较好的实际应用价值.To address the difficulties faced by educators in monitoring and evaluating learners’diverse online learning behaviors,a MLP-Bagging ensemble classification model with an embedding layer is proposed to analyze and distinguish learners’online learning behaviors.Considering the online learning behavior and individual characteristics of learners,an online learning behavior model is constructed from five aspects:learning readiness behavior,knowledge acquisition behavior,interactive learning behavior,knowledge consolidation behavior,and auxiliary features.The MLP-Bagging ensemble classification model is used to classify and distinguish learners.The experimental results show that the constructed learning model can model the learning behavior of online learners in line with reality.Adding auxiliary features is conducive to in-depth analysis and guidance of various types of learners’online learning behavior,and adding an embedding layer in the classification model can effectively overcome the data redundancy and error defects caused by label encoding,thereby achieving better classification results.Compared with other classification models,the Bagging ensemble model that integrates multiple MLP classifiers can reduce the variance of a single MLP classifier,with a classification accuracy of 98.72%,and has good practical application value.

关 键 词：在线学习行为 学习者分类 嵌入层 MLP神经网络 Bagging集成学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]