Information Gap and Feasibility-Aware Inference in Binomial Logistic Mixtures
A new paper published on arXiv explores the information gap in binomial logistic mixtures, specifically the difference between detecting mixture structure and recovering labels. The research identifies a "detectable-but-unrecoverable" regime where statistical criteria like BIC can identify components, but the associated labels remain uninformative. To address this, the paper proposes two feasibility-aware inference procedures designed to improve label recovery and posterior probability calibration. AI