Design de plan expérimental optimal : vers une approche basée sur la logique floue

We present in this article the foundation of a method using fuzzy logic for experimental plan design under constraints and with few data. The goal is to provide experimenters with an interpretable / explicable algorithm, allowing them to sample optimally. We detail the different steps of our algorithm (generation of the rule system, fuzzy inference,...), the results obtained, and the comparison with Bayesian optimization.