Multiclass Calibration Assessment and Recalibration of Probability Predictions via the Linear Log Odds Calibration Function
Amy Vennos, Xin Xing, Christopher T. Franck · 2026
Read Multiclass Calibration Assessment and Recalibration of Probability Predictions via the Linear Log Odds Calibration Function by Amy Vennos, Xin Xing, Christopher T. Franck (2026) 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-2602-18573.txt.
Description
Amy Vennos and two co-authors make probability predictions trustworthy. Their 2026 paper notes machine-made probability predictions are vital in modern classification such as image classification. A model counts as calibrated when predictions match observed rates. Existing multicategory recalibration methods are weak: they compare models rather than testing one model, require under-the-hood model access, and produce output hard for analysts.
The team proposes Multicategory Linear Log Odds recalibration, which includes a likelihood ratio test for calibration, needs no under-the-hood access and thus applies widely, and can be easy to read. They demonstrate effectiveness through simulations and three case studies: image classification with convolutional networks, obesity analysis with random forests, and ecology with regression models. Compared with four recalibration techniques using their test and Expected Calibration Error, the method works well alone and with others.
Catalog record
| CREATOR | Amy Vennos, Xin Xing, Christopher T. Franck |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-18573