User´s manual operation for vectors in interactive differential evolution for optimizing fragrance composition

Fragrance is one of important media types that have effect on human psycho-physiological states. Using fragrance suited to each user´s preference and feelings seems important, however, adjustment of fragrance composition is difficult for most of users. Interactive Evolutionary Computation (IEC) is known as an effective approach optimizing media contents, and IECs for optimizing fragrance composition have been proposed. To enhance the optimization ability of IEC, the author proposed an IEC optimizing fragrance composition with user´s operation by referring to previous IECs that accept user´s active intervention as operations on solution candidate. In the proposed IEC, Tabu Search was employed as evolutionary algorithm. Based on these previous IEC studies, this study proposes a new IEC for optimizing fragrance composition with user´s manual operation. Differential Evolution (DE) was employed as evolutionary algorithm. DE´s vector was corresponded to values of intensity of aroma sources. To construct a system based on the proposed method, paired comparison based Interactive Differential Evolution (IDE) was employed. While the user just evaluates the presented fragrances with paired comparison in the conventional IDE, the user not only evaluates the fragrance but also operates the composition of selected fragrance of DE´s vector.