Rethinking Trust Region Bayesian Optimization in High Dimensions

An arXiv RSS batch of new stat-ML papers (Apr 28, 2026) covering methods and theory for high-dimensional optimization, clustering, causal inference, kernel and spectral learning, neural-network representations, online second-order inference, probabilistic graphical models for Bayesian inversion, semi-supervised classification, regularized optimal transport, and score-based reduced-order stochastic modeling. Highlights include AdaScale-TuRBO (lengthscale scaling for robust trust-region Bayesian optimization), Turtle Shell (mixture-based discriminative clustering with automatic component number選