Performance Evaluation of Interactive Evolutionary Computation Applying Gaze Information

Abstract

We present an Interactive Evolutionary Computation (IEC) system that applies user gaze information. Historically, IEC systems have faced the problem of heavy user evaluation loads. Therefore, to solve this issue, we apply a user gaze information approach to solve such issues. When user gaze information is applied to the evaluation of candidate solutions, IEC systems can obtain user evaluation information while they view multiple candidate solutions. In this paper, we verify the effectiveness of the eye tracking IEC system using evaluation experiments with real users. In the experiment, we use a normal IEC system as a comparison method where users evaluate candidate solutions by 10-stage evaluation manually. The experimental results show that the eye tracking IEC method can generate solutions that offer results equivalent to the compared system.

9th Kansei Engineering and Emotion Research Conference. KEER2022 : Barcelona. 6-8 September 2022 : proceedings

“…Figure 2 shows the gene coding of the running shoe design and part designs. The running shoe designs are identical to those in our earlier study (Takenouchi & Tokumaru, 2019a). The design consists of five parts: base, logo, shoelace, sole, and toe.…”

“…This method is effective for product recommendation on the Internet because it considers users' preferences multidirectionally. To apply the multimodal retrieval of candidate solutions to an IEC method, our previous studies used a parallel interactive tabu search (PITS) algorithm with multiple tabu search (TS) retrievals and a hybrid GA-TS algorithm (Domae et al, 2013;Takenouchi et al, 2019a). However, the PITS algorithm generates multiple TS retrievals simultaneously, retaining those with a higher user evaluation and deleting those with a lower user evaluation.…”

Interactive Evolutionary Computation System with User Gaze Information

“…We use running shoes designs as IEC evaluation objects similar to our previous system [16]. Figure 3 shows the gene coding of the running shoes design system, which consists of the following five parts: sole, base, toe, line, and shoelace.…”

“…Previously, we applied user gaze information for evaluating candidate solutions of IEC and verifying the effectiveness of the proposed system by user experiments [16]. In this study, we investigate more effectiveness of the method with gaze information in terms of the evolutionary performance of candidate solutions.…”