The paper studies how coarsening a calibrated probability vector—by converting it to a hard label such as an argmax or a confidence threshold—affects the estimation of a treatment effect vector τ in a partially linear regression setting. It shows that the plug‑in estimator converges to a distorted version Δτ, where the distortion operator ΔΔ depends on the regression of the discarded part of the score on the retained part. The authors derive how this distortion drives coverage loss of Wald confidence intervals, provide estimable formulas for the bias and coverage, and demonstrate severe loss in simulations and real‑data audits.
arXiv:2608.21262v1 Announce Type: cross Abstract: Many machine-learning systems set a threshold at a quantile of a calibration set: conformal predictors that promise 90% coverage by drawing their cut...
arXiv:2608. 19376v1 Announce Type: cross Abstract: Split-conformal prediction provides marginal coverage under exchangeability and is increasingly used as an abstention layer for zero-shot vision-language models (VLMs).
arXiv:2609.38917v1 Announce Type: new Abstract: A classifier's conditional accuracy can change while its confidence distribution stays exactly the same. We study the worst-case movement of the reliab...
arXiv:2410.18321v3 Announce Type: replace Abstract: Confidence calibration matters wherever a classifier's probabilities, not just its labels, are consumed downstream. We study Focal Calibration Loss...
arXiv:2608. 01032v1 Announce Type: new Abstract: Training error is what we can observe on a training set; test error is the quantity we actually care about.
arXiv:2606. 19491v1 Announce Type: new Abstract: Pretrained transformers sit near singular minima of the loss, where the Fisher information metric degenerates along dead directions: directions in parameter space along which the directional Fisher vanishes.
arXiv:2602. 21160v3 Announce Type: replace-cross Abstract: In safety-critical classification, the cost of failure is often asymmetric, yet Bayesian deep learning summarises epistemic uncertainty with a single scalar, mutual information (MI), that cannot distinguish whether a model's ignorance involves a benign or safety-critical class.
The paper introduces a reference‑free instrument that, from a single fit and without an oracle, can detect whether a hybrid PDE‑parameter estimator’s assumed operator is misspecified and distinguish this from mere parameter unidentifiability. In a self‑adjoint parabolic inverse problem, the proposed information‑matrix statistic correctly identifies misspecification with low false‑positive rates, while remaining silent when the design is correctly specified but non‑identifiable. The study demonstrates that conventional accuracy checks can miss significant operator errors, and it maps out the instrument’s blind spots and conditions under which its guarantees hold.
arXiv:2608.29705v1 Announce Type: cross Abstract: Feed-forward 3D reconstruction models emit a per-pixel confidence that downstream systems read as a reliability signal. It is trained as a loss weigh...
arXiv:2607. 05806v1 Announce Type: new Abstract: Training data for machine learning is routinely collected by a selection process the model never sees: loans are observed only when granted, outcomes only when a test was ordered.
Split-conformal prediction provides marginal coverage under exchangeability and is increasingly used as an abstention layer for zero-shot vision-language models (VLMs). We audit this practice under de...
arXiv:2608. 07795v1 Announce Type: cross Abstract: Prediction intervals for multi-modal regression with tabular variables, text, images, or other input sources are difficult to calibrate when those sources disagree or one is missing.