Decentralized Conformal Novelty Detection via Quantized Model Exchange

Collection of recent stat/ML papers (arXiv) presenting methodological advances across decentralized and privacy-aware model exchange, conformal and distributional prediction, optimal-transport-based causal metrics, scalable geometry-aware distances, theory for high-dimensional ICA and Transformers, decentralized fixed-point solvers, and label-efficient inference for healthcare. Key contributions include: a quantized-model-exchange framework for decentralized novelty detection with finite-sample global FDR guarantees; Active Multiple-Prediction-Powered Inference (AM-PPI) for cost-aware, label‑s