基于PCA-LVQ神经网络的教学质量评价模型研究

Teaching Quality Assessment Model Based on PCA-LVQ Neural Network

为了进一步提高教学质量评价准确率,提出一种基于主成分分析(PCA)和学习矢量量化神经网络(LVQ)相结合的教学质量评价模型.使用层次分析法(AHP)建立教学质量评价体系,再用主成分分析提取初始评价指标体系的特征信息,将经过降维处理后的特征信息输入到LVQ神经网络,并对网络模型进行训练和泛化能力测试.实验结果表明,与单一的LVQ和BP神经网络相比,PCA-LVQ网络模型的结构更为简化,学习能力更强,收敛速度更快,评价准确率更高且泛化能力强.

To improve the accuracy of teaching quality assessment,an assessment model based on Principal ComponentAnalysis（PCA）and Learning Vector Quantization（LVQ）is proposed. A teaching quality assessment system isestablished using Analytic Hierarchy Process（AHP）,and characteristic information of indexes in the initial assessmentsystem is extracted by PCA. The characteristic information is put into the LVQ neural network after dimension reduction.The network model is trained and its generalization ability is also tested. The experiment result shows that,comparedwith simple LVQ neural network and simple BP neural network,PCA-LVQ network model has simpler networkstructure,better learning ability,faster convergence speed,higher accuracy and better generalization ability.