Active-learning guided refinement shrinks the design space by 45-50% while retaining over 99% of the Pareto-relevant hypervolume, and warm-started multi-objective Bayesian optimization finds high-value candidates faster.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Active Learning Guided Design Space Refinement for Scalable Multi-Objective Bayesian Optimization in Materials Discovery
Active-learning guided refinement shrinks the design space by 45-50% while retaining over 99% of the Pareto-relevant hypervolume, and warm-started multi-objective Bayesian optimization finds high-value candidates faster.