Researches on Emergent Design by means of Interactive Evolutionary Computation and Genetic Programming

点击翻译按钮获取中文摘要

In this research, we have used Genetic Programming (GP) to establish the effective emergent design system by means of Interactive Evolutionary Computation (IEC).Conventional optimization methods optimize target systems based on evaluation or cost functions and have been used for most of system optimization and parameter tuning. According to the increase of the importance of human factors, however, the cases that the performance of system design must be evaluated based on human subjective evaluations and domain knowledge came to increase. These cases include artistic systems generating preferable computer graphics (CG) or music, hearing-aid fitting for maximizing hearing satisfaction of its users, robots whose behaviors let our human feel comfortable, and others. IEC is a technique that optimizes target systems based on human evaluation. The number of IEC research has increased since 1990's, and it is widely used in artistic field, engineering field, and education and other fields nowadays. Not only IEC application research but also IEC interface research to reduce IEC users' fatigue problem increases, too.To solve the above difficulties, in this research, we have realized GP-based emergent design framework. The effectiveness of our approach has been demonstrated by several applications, e.g., analog circuit design, musical composition, evolutionary robotics, agent designs in WWW space, and multi-agent cooperation learning.