시각의 환원 원리를 적용한 인공지능으로 아름다움 재현하는 구조에 관한 연구

Abstract

The beauty of visual stimuli is a complex whole composed of many factors. These wholes can be divided into individual cognitive elements, and the brain can process information in this mechanism. We only perceive beauty as a whole without knowing what the components and structures are, and as a result, it is challenging to reproduce beauty through these. But the brain has evolved over a long period to have the intelligence to reduce the whole to multiple levels, and neuroscience has shown the process well. In addition, recently, active artificial intelligence has imitated it and succeeded in reproducing beauty.
 In this study, visual stimuli were reduced and organized into six levels of intelligence through prior research in cognitive neuroscience. First, the retina converts light into electrical pixel data, which is transmitted to the Lateral Geniculate Nucleus (LGN). Information arriving from the Lateral Geniculate Nucleus to the V1 area is rapidly transferred to the cognitive areas of the brain through the IFOF channel, causing visual attention. In order to respond quickly to the state of the world, a map of spatial visual coordinates is formed in the sensory and perceptual states as a mechanism for attending to objects. Second, primary visual processing reduced the attended visual stimuli to Feature, and the secondary visual cortex was reduced by processing visual information in Element, Shape, and From. Next, the structure of the connections between the visually reduced elements and the affective domain was examined, and it was found that the domain processing the From had a higher density of opioid receptors that elicited positive affect. Then, these visually reductive elements are converted into vocabularies, and a proposition is generated to translate them into Modellings. The generated vocabulary and propositions are used to create a beautiful design using the artificial intelligence Midjourney. Finally, the research on the goodwill of the beautiful shape is carried out, which further verifies the feasibility of this study.
 This study is based on cognitive neuroscience, which reduces beauty as a whole to visual stimuli. The possibility of recognizing the design is verified by the beautiful reproduction of artificial intelligence. That is, an attempt is made to transfer the process of understanding and to reproduce the structure of beauty from the feeling of beauty to the category of recognition and to solve the difficulties encountered in producing beauty.