摘要
为解决传统学习模式自动识别方法中卷积核自身几何设计有限导致的识别精度较低的问题,提出基于可变形卷积网络的思政课堂移动学习模式自动识别。在经典网络中加入与学生头部运动情况相关的偏移矢量;利用可变形卷积层、池化层与全连接层构成可变形卷积网络模型;经过获取初始权值分布状况、误差计算等步骤完成网络模型训练;引入激活函数,调节神经元数量,完成学习模式自动识别。仿真实验证明,所提方法具有较高的自动识别精度与良好的数据处理能力。
In order to solve the problem of low recognition accuracy caused by the limited geometric design of convolution kernel in traditional automatic learning pattern recognition method,an automatic recognition method of mobile learning pattern in ideological and political classroom based on deformable convolution network is proposed.The offset vector related to the movement of students’head is added into the classic network,and the deformable convolution is formed by using deformable convolution layer,pooling layer and full connection layer.After obtaining the initial weight distribution and error calculation,the network model training is completed,and the activation function is introduced to adjust the number of neurons to complete the automatic recognition of learning patterns.Simulation results show that the proposed method has high automatic recognition accuracy and good data processing ability.
作者
杜悦
刘婷
DU Yue;LIU Ting(Nursing School of Harbin Medical University Daqing Campus,Daqing 163319 China)
出处
《自动化技术与应用》
2022年第8期101-105,共5页
Techniques of Automation and Applications
关键词
可变形卷积网络
思政移动课堂
学习模式
自动识别
激活函数
deformable convolution network
ideological and political mobile classroom
learning mode
automatic recognition
activation function
作者简介
杜悦(1988-),女,硕士研究生,助教,研究方向:马克思主义中国化研究。
