New research proposes budget-optimal data collection for machine learning

A new research paper published on arXiv details a method for optimal data collection from multiple, heterogeneous sources under a fixed budget. The proposed approach maximizes effective sample size by considering the costs associated with different data sources and their group compositions, using a sampling plan that minimizes $\chi^2$-divergence. This method is paired with a post-stratification estimator to achieve budgeted minimax optimal risk for estimating population and group-conditional means, and can be extended to prediction problems. AI

RANK_REASON Research paper published on arXiv detailing a new methodology. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• arXiv
• D_{\\chi^2}(qcolorbox{white}{white}{||}\\overline{p})
• machine learning
• Michael E Harding

• paper
• other

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