SCOPE-FE: Structured Control of Operator and Pairwise Exploration for Feature Engineering

Batch of new arXiv/stat-ml submissions covering advances in automated feature engineering (SCOPE-FE: operator probing + feature clustering + reliability scoring for scalable candidate reduction), causal inference methods (tree-based discretization with ILP matching; Bayesian X-Learner for calibrated posteriors under heavy-tailed outcomes), semi-supervised inference via mixture-of-experts for prediction-powered inference, value-aware revenue-focused recommendation with high-dimensional similarity measures, robust/confidence-interval theory in Efron’s two-groups model with a Fourier-based certif