Running shoes design system with artificial bee colony method using gaze information

Abstract

To retrieve multimodal candidate solutions for real users, we investigated the effectiveness of an interactive evolutionary computation (IEC) method with an artificial bee colony (ABC) algorithm. Using three types of bees (i.e., employed, onlooker, and scout bees), the ABC algorithm retrieves various candidate solutions. Our previous study showed the effectiveness of the IEC with the ABC algorithm while looking at various practical IEC parameters from a numerical simulation using a pseudo-user that imitates user preferences. The results showed that the IEC with the ABC algorithm could retrieve more multimodal candidates than the interactive genetic algorithm (IGA), the previous chief method of IECs. However, we did not examine the effectiveness of the IEC with the ABC algorithm for real users. In this study, we performed experiments to examine the effectiveness of the IEC with the ABC algorithm for real users using running shoe designs as an evaluation object. The investigations compared multimodal candidate solutions using the IGA method as a comparison tool, retrieving the performance of both methods. To evaluate candidates, we employed user gaze information to reduce user evaluation loads. The results showed that the evaluation time for evaluating candidates of the IEC with the ABC algorithm was shorter than that of the IGA method. Moreover, we confirmed that the IEC with the ABC algorithm could retrieve more multimodal candidate solutions than the IGA method.