Generalized Neural Distributional Regression

This collection of arXiv preprints (July 17, 2026) covers advances in statistical machine learning, probabilistic modelling, and scalable inference. Highlights include: Generalized Neural Distributional Regression (GNDR) — a semi‑parametric two‑step approach that embeds neural networks into classical distributional parameter spaces and yields analytical Fisher information and observation‑specific confidence/tolerance intervals (implemented in the open‑source package thetaflow); extension of operator‑informed Gaussian processes to complex Helmholtz wavefields with application to in vivo brain M