A proposal for distributed interactive differential evolution
Makoto Fukumoto, Kota Nomura · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2018
Interactive Evolutionary Computation (IEC) is used for creating media contents suited to each user's feelings. To obtain media contents commonly suited to multiple users, distributed approach was employed in previous IECs based on Genetic Algorithm. This study proposes Distributed Interactive Differential Evolution (DIDE) by employing Differential Evolution (DE). Based on each user's subjective selection and scoring with Interactive DE, DIDE exchanges good solutions between users. Through the exchange, it is expected that the good solution for all users is obtained. A fundamental experiment was conducted with a simple DIDE system creating sign sounds for two users. Distance between DE vectors obtained by two users and fitness values were investigated.