【作 者】：
徐振国， 张冠文， 孟祥增， 党同桐， 孔 玺
【关 键 词 】：
深度学习； 学习者情感； 情感识别； 智慧学习环境； 情感交互； 卷积神经网络
【栏 目】：
学习环境与资源
【中文摘要】：
情感能够影响和调节学习者的注意、记忆、思维等认知活动，学习者情感的准确识别是构建智慧学习环境中和谐情感交互的基础，更是判断学习者学习状态的重要手段。传统学习者情感识别方法存在识别率低、算法复杂、鲁棒性差等问题，并且容易丢失面部表情特征的关键信息。文章提出一种基于卷积神经网络的学习者情感识别方法，该网络包括3个卷积层、3个池化层和1个全连接层。然后在自主搭建的大规模学习者情感数据库中进行了训练和实验，实验结果表明该方法能够快速、准确的识别学习者情感。未来，该方法可应用到智慧学习环境建设中，为完善学习者模型、实现情感交互、挖掘学习行为等提供技术支撑。
【英文摘要】：
Learners' attention, memory, thinking and other cognitive activities can be affected by their own emotions. Accurate identification of learners' emotions is the basis of building harmonious emotional interaction in smart learning environment, and it is also an important means to judge learners' learning status. Traditional methods for learners' emotion recognition still have some problems such as low recognition rate, complex algorithm, poor robustness, and losing the key information of facial features easily. This paper proposes a learner emotion recognition method based on convolutional neural network, which consists of three convolutional layers, three pooling layers and one full connection layer. Then the training and experiments are carried out on the large-scale learner emotion database that is independently built. The experimental results show that this method can identify learners' emotions quickly and accurately. In the future, this method can be applied to the construction of smart learning environment to provide technical support for improving learner model, realizing emotional interaction and mining learning behavior.