New research explores admissibility geometries in predictive inference

A new research paper titled "Bayes with No Shame: Admissibility Geometries of Predictive Inference" has been published on arXiv. The paper, authored by Daniel Zantedeschi, explores four criterion-relative geometries in modern predictive systems, including Blackwell risk dominance, anytime-valid admissibility, fixed-level marginal coverage, and choice-based approachability. It establishes pairwise non-nesting across these different predictive system types and analyzes various coherence notions, such as posterior predictive means as martingales and the equivalence of anytime-valid admissibility to the nonnegative martingale property in certain models. The research also details how coverage admissibility is certified by exchangeability ranks and boundary-feasibility by Cesaro steering, organizing these paradigms within a constrained-Bayes design schema. AI

IMPACT This research contributes to the theoretical foundations of predictive systems, potentially influencing future developments in machine learning algorithms.

RANK_REASON The item is a research paper published on arXiv detailing theoretical advancements in predictive inference. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• alphaXiv
• arXiv
• Bayes with No Shame: Admissibility Geometries of Predictive Inference
• Blackwell
• Capparis
• CatalyzeX
• Core Recommendations for Antifungal Stewardship: A Statement of the Mycoses Study Group Education and Research Consortium
• DagsHub
• Daniel Zantedeschi
• Gotit.pub
• Hugging Face
• ScienceCast

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →