IoT-Based Behavioral Analysis Model for Smart Learning

The role of education is of paramount importance in influencing the cognitive development of future generations. This paper presents a novel IoT-based Behavioural Analysis (IoT -BA) Model for Smart Learning integrated with neural network technology. The Internet of Things (IoT) enables educational institutions to perform management tasks more quickly and risk- freely by analyzing organized data and acquiring insights from it to inform decisions. The significance of teaching college students in the self-directed application of effective learning strategies has evolved with various trending digital twin frameworks. The proposed IoT -BA model can collect the motion features and facial expressions of the students from the classroom video while attending classes through self-learning applications. The collected images can then be preprocessed using augmentation and a local binary pattern approach. This deep-preprocessed data has been passed to the deep learning algorithm convolutional neural network (CNN) that is trained with the preprocessed data for classification. Spatiotemporal feature extraction and selection from the preprocessed data can be performed using the preprocessing algorithm principal component analysis (PCA). The student's attention level classification for the IoT -BA uses the CNN algorithm. This experimental research uses the real-time data collected by setting the classroom environment for model training and validation. The CNN model has been trained with the data collected from the environment using the simulation platform TensorFlow and performs five-cross-validation with the realistic data. The IoT -BA model can predict the student's attention level in the classroom based on the collected features from the classroom video. The simulation results of the CNN -based IoT -BA model showed better performance in terms of 96.42% classification accuracy and a minimal error rate of 1.82%.