Face Emotion Pattern Analysis of Korean Depending on Persons and Environments Using DCGAN

Abstract: Many studies examine the effects of a knowledge management, self-directed learning, emotion intelligence, and creative performance. It means that a very closed link between emotions and learning, and emotion give influence or their impact on the learning process. Adults as well as children can be influenced in their ability to direct their learning process by emotions. Emotion functions help one to remember their study results and at the same time distract them from the learning topic. That is, emotion results present different perspectives of negative and positive emotions in learning. Currently, online education is increasing and important method as one of education areas. However, we do not have an effective result because of online system. This paper suggests method to analyze efficiency of online education through the analysis of face emotion patters Face emotion patterns is different depending on race, person, and environment. First, this paper provides the results of Korean facial emotion pattern analysis to use for effective results analysis of online education. The facial emotional expression pattern, such as happiness, anger, sadness, embarrassment, injury neutral, and pleasure, is different depending on the person and country, and the results of its recognition accuracy are different depending on the learning method and structure (DCGAN, Cycle GAN, PixelGAN, DiscoGAN, StyleGAN, transfer learning). This paper analyses Korean facial emotional data created by the agency for AI infra, using DCGAN, and provides the analysis results of facial emotional patterns for another user's easing use.

Keywords: deep learning, DCGAN, face emotion, online education, Korean face emotion.