Deep reinforcement learning-based Smart Education System with wireless sensor network virtualization

Create a smart classroom that uses many sensors, such as webcams, to monitor each student's progress and leverages deep reinforcement learning (DRL) algorithms to provide personalized instruction based on each student's unique sensing data. Smart learning suggestion systems also analyze students' pulse rates, quiz scores, and expressions to deduce their current levels of comprehension. They then use DRL to provide suggestions for pupils based on where they are in their studies. It is beneficial to employ virtual reality in the classroom. Software virtualization technologies are now the main focus of IT and systems management publications. Virtualization technology might make service upkeep simpler. Changes to the classroom setting, the teaching strategy, and individual student conditions may drastically alter the efficacy of a conventional classroom. A smarter way to manage the classroom that takes context into account based on wireless sensor network (WSN) technology utilizing a virtualization process with DRL (WV-DRL) is suggested and implemented in this research to improve learning efficiency. Wireless sensors monitor students and classroom conditions, sending data to a server-based management system. The system provides feedback to both students and instructors, as well as wireless sensor-controlled classroom equipment. It uses virtualization technology to make the most of expensive classroom equipment and ensure its reliability. Smart classrooms with several sensors may model a Markov decision process to provide personalized learning suggestions. Preliminary findings indicate that our intelligent learning recommendation system is effective. This research provides useful guidelines for doing so in the future, paving the way for developing an intelligent learning environment that facilitates higher levels of personalization. The experimental results show the proposed WV-DRL to achieve an accuracy ratio of 90.2%, recall ratio of 92.3%, score query 4.7%, the f-score ratio of 96.5%, and precision ratio of 95.2 % compared to other methods.