关键词: 进化认知, 主成分分析, 演化算法, 材料质感, 偏好学习

Abstract: To enhance the product identification by material design and to improve the customer's satisfaction,an evolutionary cognition algorithm of material texture preference based on gene expression programming was proposed.By considering the influence of product texture's visual sense,tactile physical quantity and visual psychological capability on product preference comprehensively,124 floor tiles were measured in terms of color,glossiness and roughness.60 subjects scored these 124 floor tiles in terms of texture and preference subjectively.After dimensionless treatment and principal component analysis,four components were extracted as independent variables while the product preference as dependent variable,and the cognitive relationship between product material texture aspects and user preference was transformed into a modeling problem of complex functions.The results showed that this gene expression programming had obvious advantages over neutral network and support vector machines in terms of prediction accuracy and robustness on solving product material texture cognition problem.A floor preference assessment system was prototyped based on this proposed algorithm and the effectiveness was proved in practical application.

Key words: evolutionary cognition, principal component analysis, evolutionary algorithm, material texture, preference learning