Uniform convergence of the smooth calibration error and its relationship with functional gradient
Futoshi Futami, Atsushi Nitanda · 2025
Read Uniform convergence of the smooth calibration error and its relationship with functional gradient by Futoshi Futami, Atsushi Nitanda (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-19396.txt.
Description
Futoshi Futami and Atsushi Nitanda bound the calibration error of probability forecasts. Their 2025 paper studies the smooth calibration error, a gentle yardstick for honest odds. Forecasts in risky fields must be right and honest at once. Theory for achieving both has lagged behind practice.
The smooth error splits into a training term plus a generalization gap. The loss's functional gradient, they prove, can hold the training term down. Three algorithm families are then audited: boosted trees, kernel boosting, and small neural nets. For each, conditions deliver accuracy and calibration together. Designers gain rules for building forecasts that mean what they say.
Catalog record
| CREATOR | Futoshi Futami, Atsushi Nitanda |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 148 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-19396