Artificial Intelligence System Based on Visual Communication Using Convolutional Auto-Encoder and Convolutional Neural Network

The development of intelligent technology had specific impact and enabled virtuality-reality interaction in field of education. Specifically, the Digital Twins (DT) symbiosis of virtual reality, real-time interaction, and solid virtual simulation. Also, it is an extended application to the education field. This research optimizes Visual Communication (Viscom) course design based on Artificial Intelligence in Deep Learning (DL). Initially, 310-questionnaire survey was distributed randomly to various types of students in July 2021. The numerical feature values are standardized to a same interval for comparison using min-max normalization. Then, a hybrid system for Viscom using Convolutional Auto-Encoder (CAE) and Convolutional Neural Network (CNN). CAE is trained to decrease the dimensionality of input data and CNN is used to represent DL for Viscom course. After implementing DL into Viscom course, outcomes indicate that student’s interest in learning and satisfaction with practical knowledge is improved. Moreover, satisfaction with theoretical learning before practical courses is minimized and Viscom course teaching effect is increased.