New CAP method optimizes black-box tasks with sparse historical data

Researchers have developed a new method called CAP (Candidate Acquisition Policy) to improve black-box optimization when experimental evaluations are costly and limited. CAP models shared structures and task-specific variations from historical data to learn a query policy. This policy is then used to make adaptive decisions for new tasks, even with only a few sparse histories. Experiments across synthetic task families and engineering benchmarks like BOPTEST and SU2 showed CAP achieved superior optimization performance compared to other methods under fixed evaluation budgets. AI

IMPACT This method could lead to more efficient experimental design in fields where evaluations are costly, such as engineering and materials science.

RANK_REASON The cluster contains a research paper detailing a new method for optimization. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• BOPTEST
• CatalyzeX
• DagsHub
• Hugging Face
• IArxiv

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →