Applying deep learning to wireless sensor networks for monitoring students’ emotion in high schools

Abstract

Vietnamese students are facing significant academic pressure due to societal and familial expectations, which leads to an unfavorable learning environment. We aim to employ a temporary spatial-temporal stress monitoring system. Using Wireless Sensor Network (WSN) technology, it collects data on students’ emotional states and incorporates a prediction model, “Reduce Students’ Stress in School” (R3 S), to detect students’ emotional states across school premises. The integration of R3 S and WSN is conducted in three stages. Initially, sensor nodes are deployed in schools to collect emotional data. Subsequently, we introduce a novel hybrid model combining a one-dimensional Convolutional Neural Network with Long Short-Term Memory networks (1D-CNN-LSTM) to generate a predictive emotional map. This model’s performance, evaluated using RMSE and MAE metrics, shows exceptional precision compared to other LSTM models. When predicting the “stress” condition, the R3 S model achieved a Mean Absolute Error (MAE) of 10.30 and a Root Mean Square Error (RMSE) of 0.041. Lastly, we generate a comprehensive map of cumulative emotional conditions, serving as a guide for school counselors. This map aids in fostering a healthy, conducive learning environment.