Virtual Reality and 6G Based Smart Classroom Teaching Using Artificial Intelligence

Shuai Liu

DOI: https://doi.org/10.1007/s11277-024-11223-y

IF: 2.017

2024-06-19

Wireless Personal Communications

Abstract:With the rapid development of the big data era, deep reinforcement learning has gradually become a research hotspot in artificial intelligence. This study focuses on the creative use of deep reinforcement learning (DRL) in educational technology, particularly by incorporating 6G and virtual reality (VR) into smart classroom environments. This research aims to improve AI-driven educational frameworks using a novel symmetric data improvement technique with a symmetric Deep Q-Network (SDQN) model. The SDQN model adapts the learning process to the requirements of virtual smart classrooms by improving neural network parameters using the gradient descent technique and mean square error loss function. Our findings show a notable improvement in the algorithm performance, mainly when the model processes approximately 70k frames, the distance between the curves begins to increase, and the average score of the symmetric deep Q-network model is always higher than the deep Q-network model. This indicates that the former performs better. When the frame number is 98k, the average score of the symmetric depth Q network is the highest, at 97.6. After 1000 iterations of training, the symmetric deep Q network model achieved a training prediction accuracy of 99.5% and a testing prediction accuracy of 85.6%, while the training accuracy of the deep Q network model was 97.4% and the testing accuracy was 85.7%. The performance of a symmetric deep Q network is superior to previous data augmentation methods in the context of smart classrooms, showing that it has a significant chance to reduce the costs of developing educational content while improving learning outcomes through immersive, AI-enabled virtual environments.

telecommunications