Emotion Recognition System for Emotion Analysis and Decision Making of Students during Class Sessions: Integration of Real Time Process and Fuzzy Sets in Educational Psychology
Abstract
Emotions are important for the cognitive engagement, learning efficiency, and decision making of students in the classroom session. Accurate and real-time recognition of the student's emotional states can assist adaptive teaching strategies and improve the learning outcomes. This paper proposes to develop a real-time emotion recognition system that combines deep learning-based facial expression analysis with the fuzzy set theory to model the emotion uncertainty and facilitate the pedagogically driven decision making. Facial features are extracted with the convolutional neural network and fuzzy membership functions and fuzzy rules based on educational psychology are used to model overlapping emotional states and make adaptive choices with respect to the instruction. Experimental evaluation shows that the proposed framework can obtain better accuracy in recognising emotions, better decision-making performance and lower response time compared with traditional machine learning, standalone deep learning, and rule-based methods. The results confirm the usefulness of combining real-time processing and fuzzy logic in educational systems that will become emotion aware and that will allow responsive, interpretable, and student-centered classrooms.