[1M2-CC] Synergy among Machine Learning, Computational Optimization, and Human Awareness / General Session – A

[1M2-CC-03] Acquiring Consensus Solutions by Multi-human-agent-based Evolutionary Computation

Hironao Sakamoto¹, Koutaro Nakamoto¹, Kei Ohnishi¹ (1. Kyushu Institute of Technology)

The paper proposes a system to acquire consensus solutions among people to a multiobjective problem which does not require direct communication among the people. It relies on multi-human-agent-based evolutionary computation (Mhab-EC) with a mechanism for making solutions of agents converged. In the system, people prepare their own agents who can be tuned by themselves, and then submit their agents to a simulation of MhabEC and acquire a convergence solution from the simulation. Then, examining the simulation result, they tune their agents again to be closer to their desired solutions and submit the agents to the simulation and obtain a convergence solution. The mechanism repeats this procedure pre-determined times and outputs a convergence solution of the final simulation as the consensus solution. In basic evaluation of the system, we focus on if the system can make solution converged and show that it can indeed do it.