Fig. 2: Workflow.

Framework of the proposed multi-objective Bayesian active learning with included experimental data uncertainty, for designing super high-strength and high-ductility lead-free solder alloys. a Dataset with experimental uncertainty. b Two developed Gaussian Process Regression (GPR) models, one for strength and one for elongation. c Multi-objective Bayesian sampling based on the acquisition-function-modified objective space. d Iteration results with experimental feedback. e Experimental characterization and mechanism exploration.