摘要
学习情绪智能感知是学习科学的研究前沿。针对孤独症儿童情绪表达频率低、特征异常、数据集缺少和识别率低等问题,文章采用真实情景教学法,建立孤独症儿童学习情绪数据集,构建Attention-AlexNet优化模型,并将此模型与AlexNet基础模型进行情绪智能感知的对比实验,结果表明:孤独症儿童面部表情数据集具有可用性和准确性;Attention-AlexNet优化模型感知孤独症儿童情绪的准确率高于AlexNet基础模型,其对五种情绪识别的准确率由高到低依次为开心、惊讶、厌烦、生气和平静。文章推进了智能学习科学中深度学习数据集的建立和情绪智能感知的理论研究,可为孤独症儿童教育干预提供技术支撑。
The intelligent perception of learning emotion is the research frontier of learning science.Aimed at the problems of low frequency of emotion expression,abnormal characteristics,lack of data set,and low recognition accuracy of emotion expression for autistic children,this paper established the data set of learning emotion of autistic children using real scene teaching method,constructed the Attention-AlexNet optimization model,and carried out a comparative experiment of the intelligent perception of emotion between this model and the AlexNet basic model.The results showed that the facial expression data set of autistic children was available and accurate,and the accuracy of the Attention-AlexNet optimization model in emotion perception for autistic children was higher than the AlexNet basic model,and the accuracy rate of the Attention-AlexNet model for the recognition of five emotions from high to low was happy,surprise,disgust,angry,and calmness.This paper advanced the construction of deep learning data set and the theoretical research of emotionin intelligent perception in the intelligent learning science,and could provide technical support for the educational intervention for autistic children.
作者
王永固
张晨焘
许家奇
李晓娟
WANG Yong-gu;ZHANG Chen-tao;XU Jia-qi;LI Xiao-juan(College of Educational Science and Technology,Zhejiang University of Technology,Hangzhou,Zhejiang,China 310023;College of Marxism,Zhejiang University of Finance&Economics,Hangzhou,Zhejiang,China 310018)
出处
《现代教育技术》
CSSCI
2021年第7期105-111,共7页
Modern Educational Technology
基金
国家社会科学基金项目‚残障儿童‘家庭—社会’融合型关爱长效机制构建及制度创新研究‛(项目编号:16BSH139)
浙江省社会科学规划重大项目‚‘互联网+’残疾人社会支持智慧服务构建与创新‛(项目编号:20XXJC01ZD)
教育部人文社会科学研究青年基金项目‚‘互联网+’自闭症家庭精准帮扶的协同机制与实现模式研究‛(项目编号:18YJCZH085)
浙江省高校科研基本业务费跨学科研究专项‚面向智能经济的AI+创客空间构建与服务创新‛(项目编号:GB201903003)的阶段性研究成果。
关键词
深度学习
情绪识别
孤独症儿童
卷积神经网络
智能感知
deep learning
emotion recognition
autistic children
convolutional neural networks
intelligent perception
