多级Sigmoid神经网络在本科教学评价中的应用

Application of University Teaching Evaluation by Using Multi-Level Sigmoid Neural Network

为了解决高校教师教学质量评价不全面、主观性过强等问题,提出了采用多级Sigmoid神经网络评价本科教学效果的方法.首先采用主成分分析从原始指标特征中提取有效信息,极大地减少了指标之间的相关性;其次,再通过建立类似于大脑神经突触信息神经网络处理模型,实现对本科教学效果的自动评价.实验结果表明,相对于其他评价方法,该方法具有更好的泛化性能,能提高本科教学的评价效果.

In order to improve the incomplete and subjective evaluation of university teaching quality, a new teaching quality evaluation method by using multi-level sigmoid neural network was proposed. Some effective information was ex- tracted from the original input information by using the principal component analysis to reduce the correlativity among in- put variables. Then the multi-level sigmoid network model, which simulate the information processing in human brain synapse, was established to evaluate the teaching quality automatically. The experimental results show the generalization performance and effectiveness of the proposed method.