UchUchU
stat.MLastro-ph.COastro-ph.IMcs.LG

$\texttt{NFDiag}$: High-dimensional MCMC Convergence Diagnosis with Normalizing Flows

Kutay Nazlı, Elena Sellentin

概要

Accurately assessing the convergence of Monte Carlo Markov Chains is crucial for ensuring their validity in Bayesian inference problems. However, convergence assessment in high dimensions is known to be difficult, due to most convergence diagnostics becoming insensitive beyond a few dozen parameters. Conventional convergence statistics analyze marginal distributions which in high dimensions neglects increasing levels of information from the joint and conditional probability distributions. Thus w…

出典：arXiv ／ PDF

関連するニュース

PTED: A multi-dimensional two-sample test for scientific inference and generative machine learning astro-ph.IM · 2026.09.30 Recovering Weak Signals with Normalizing Flows stat.ML · 2026.09.06 Traceable Spectral Inference via Influence Functions: Efficient Data Attribution and Error Proxies for the Ariel Mission cs.LG · 2026.08.25 Don't Cut Corners: How Training Outside the Prior Makes Simulation-Based Inference More Robust astro-ph.IM · 2026.08.13 Multiclass Classification without Labels via Posterior Simplex Geometry cs.LG · 2026.07.28 Covariance-Boosted Gaussian Processes for Spatiotemporal Irregularities stat.ML · 2026.07.25

この分野の企業を探していますか？

UchUchUは宇宙産業に関わる日本企業をデータベース化しています。参入検討・取引先探索にお使いください。

企業データベースを見る 参入ガイドを読む
← 研究動向へ戻る