As introduced on the page of human preference analysis,Person has various preference. We think it should consider to generate user’s favorite design with IEC (Interactive evolutionary computation). In the research field of evolutionary computation algorithm,Researchers have tried to search candidate solutions in multimodal and retrieve candidate solutions that satisfies multi-objective standards. Even in IEC system, such techniques should be applied, currently,we now proceed various inspections related multimodal and multi-objective retrieving using numerical simulations.

For example, we are working on hybrid search that combines genetic algorithm which is a global search method and taboo search method which is a local search method, application of Artificial Bee Colony algorithm (ABC algorithm) invented for purpose of multimodal retrieving.



By such works, for example, about user’s favorite design, the system may present designs of various tastes. But new IEC system which included multimodal retrieving will be needing from now on. Previous IEC system creates only a single design of user’s favorite.

So far, we have conceptualized an interactive evolutionary computation system using the Artificial Bee Colony algorithm and conducted various validations based on numerical simulations. The figure below illustrates the flow of this system.


As a result, we confirmed that, compared to conventional evolutionary computatsion, the proposed system can explore the solution space multimodally with a significantly smaller number of candidate solutions and generations. Moving forward, we plan to enter the phase of applying this to real systems.

Related our works
[Academic Papers] Hiroshi Takenouchi, Masataka Tokumaru, "Interactive Evolutionary Computation with Artificial Bee Colony Method for Multimodal Preferences Retrieval", International Journal of Affective Engineering, Vol.22, No.1, pp.1-10, 2023-01.