Quantification of Aesthetic Emotional Features and Machine Learning Based on Reduced Dimensional Assignment

Abstract

For the difficulty of identifying and analyzing human aesthetic and emotional factors occurred in the development of artificial intelligence (AI), the product modeling as the research object is conducted with the analysis of aesthetic emotion to try to solve the difficulty of providing effective quantitative data for machine learning. According to the idea that the mathematical methods including factor analysis, independ component analysis, principal component analysis, and other data high-dimensional feature preprocessing mathematical methods in machine learning is introduced into the digital transformation of product modeling and aesthetic emotional characteristics, the "reduced dimensional assignment method" is proposed, with its specific application demonstrated by analyzing the modeling design of a kettle body. The results show that the judgment on the new modeling design of kettle body by machine learning data that are transformed by the method in study and analyzed by computer is highly consistent with the artificial subjective judgment of the control group, which preliminarily verifies the validity of reduced dimensional assignment method, and also proves the feasibility of realizing more accurate analysis and prediction on people's aesthetic feeling of various works through machine learning.