Mean Testing under Truncation beyond Gaussian

Collection of new arXiv ML/stat papers (2026-05-05) covering: high-dimensional mean testing under arbitrary truncation and detectable bias floors; a differentially-private LASSO method that corrects anisotropy from heterogeneous covariates via Gram-based objective perturbation; scale-invariant self-normalized martingale bounds for online regression; PRCD-MAP for per-edge calibrated priors in causal discovery; MissBGM, a Bayesian generative imputation method with open-source code; distributional causal mediation with conditional generative models; a semi-supervised kernel two-sample test that利用