Metraipolis › archive › dc_da00793c668e
Calibrating your confidence requires thinking of all the ways you could be wrong — arXiv:2611.08997
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.08997
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-09T22:41:21.999Z
- SHA-256 of the archived text
6346a428954e9fca56e1cf59e64391d44cc6a7bbface36cd556551519b1d13dd- Stable URL of this page
https://metraipolis.com/documento/dc_da00793c668e
How to cite
Calibrating your confidence requires thinking of all the ways you could be wrong — arXiv:2611.08997 — arXiv (en), https://arxiv.org/abs/2611.08997 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_da00793c668e (Metraipolis archive dc_da00793c668e, sha256 6346a428954e9fca56e1cf59e64391d44cc6a7bbface36cd556551519b1d13dd)
BibTeX
@misc{dcda00793c668e,
title = {Calibrating your confidence requires thinking of all the ways you could be wrong — arXiv:2611.08997},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.08997},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_da00793c668e; sha256 6346a428954e9fca56e1cf59e64391d44cc6a7bbface36cd556551519b1d13dd},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.08997. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
A primary limitation of modern deep learning is the miscalibration of model confidence; traditional architectures fundamentally lack a mechanism to quantify their own uncertainty. Larger models and larger datasets mitigate this issue, but they do not resolve the underlying structural deficiency. Bayesian methods offer a principled uncertainty-aware learning framework through stochastic parameters. Yet, despite the established benefits, the inherent computational overhead often compromises either scalability or quality. This trade-off currently prevents the widespread adoption of Bayesian methods. This Thesis bridges the gap between high-quality and low-complexity by introducing a novel distribution on parameter space: a degenerate Gaussian whose covariance is a projection matrix aligned with the null space of the Fisher information matrix. We demonstrate the scalability of this Bayesian posterior through two numerically efficient algorithms: one for posterior sampling and another for the direct computation of predictive confidence. Empirically, we show State-Of-The-Art confidence calibration and out-of-distribution detection for large models. We provide theoretical grounding for this approach. We study the degeneracies inherent in Amari's information geometry and highlight its most relevant component: the split of parameter's perturbations into two orthogonal subspaces. From this perspective, our posterior naturally emerges as the simplest choice that captures this split, assigning unitary variance to one subspace and zero variance to the other. Moreover, through the tools of differential geometry, we utilize this split to identify and rectify a critical misalignment issue of the non-linearized Laplace Approximation. Finally, with a focus on PAC-Bayes generalization bounds, we pave the way for a deeper understanding of the learning capabilities of overparametrized models. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Machine Learning (stat.ML). Pagina del lavoro: https://arxiv.org/abs/2611.08997 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-09T22:33:47.148Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).