Autonomic Nervous Pattern Recognition of Students’ Learning States in Real Classroom Situation

The current work applied machine learning methods to analyze students' group and individual learning states in the real classroom environment. The purpose is to provide insights into the learning process for teachers and students, so that they can manage the learning process effectively. We extracted three couples of learning states in real classroom learning, i.e., information input/processing and retrieval/processing states, cognitive load matching and mismatching states, and mental fatigue and nonfatigue states. The recognition of the above three couples of learning states was regarded as five binary classification problems. We collected electrocardiogram (ECG) data from 45 college students during their classes of circuit analysis and calculated the peaks of two consecutive R waves (RR) interval series from the ECG data. For each binary classification problem, RR interval features and classifiers were compared to find the critical feature subsets and suitable classifiers for the recognition of learning states. The generalization accuracies of the machine learning models were in the range of 58.41%–82.35% on the validation sets independent of classifier training and feature selection. The results show that it is feasible to monitor students' learning states through machine learning methods.