Adaptive sampling methods via machine learning for materials screening

High-throughput virtual screening by using a combination of first-principles calculations and Bayesian optimization (BO) has attracted much attention as method for efficient material exploration. The purpose the is often to search materials whose properties meet certain target criterion, while conventional BO aims find global extremum. Some recent works use converting such motivation. On other hand, an adaptive sampling method, where acquisition function based on probability that data point achieves property within specific range, suggested previously [Kishio et al., Chemom. Intell. Lab. Syst. 127, 70 (2013)]. In this paper, we demonstrate effective exploration criteria. We conducted simulations in-house database constructed compared performance approaches. Furthermore, evaluate discuss functions extended multi-objective problems exploration, considering multiple-target simultaneously.